The Gang Truce as a Form of Violence Intervention Technical Report

July 2015

The Gang Truce as a Form of Violence Intervention

Technical Report

SolucionES Project Cooperation Agreement No AID-519-A-12-00003

Coordination Document prepared by Fundacioacuten Nacional para el Desarrollo - FUNDE Roberto Rubio Fabian PhD ndash Executive Director FUNDE Patricia Valdeacutes ndash Project Coordinator SolucionES-FUNDE Juan Meleacutendez ndash Research Technician Project SolucionES-FUNDE

Authors Charles M Katz PhD Arizona State University Luis Enrique Amaya MS

San Salvador El Salvador CA 2015

ACKNOWLEDGEMENTS

This document was made possible through the generous support of the people of the United States of America

through the United States Agency for International Development under cooperation agreement no AID-519-A-12-

00003 The contents of this document are the sole responsibility of the authors and do not necessarily reflect the

views of the United States Government The research was also funded in part by the ASU Foundation through a

generous gift by the Watts Family The opinions expressed herein are those of the authors and are not necessarily

those of the ASU Foundation or the Watts Family We would also like to acknowledge support from Dr Anthony

Harriott who provided data and contextual support from Jamaica John Maisto who assisted with data acquisition

in Honduras Eric Hedberg who provided statistical assistance and Andrew Fox Cher Stuewe-Portnoff Beto Brunn

Giuliano Perseu Juan Melendez Patricia Valdes Tim Nelson and Lidia Nuno for their helpful comments and

suggestions throughout the project

July 2015

Table of Contents

Page Executive Summary i

Section 1 Introduction 1

Section 2 Gang Truce - The Salvadorian Experience 11

Section 3 Gang Truce ndash The Jamaican Experience 44

Section 4 Gang Truce ndash The Honduran Experience 70

Section 5 Conclusions Policy Implications and Recommendations 85

References 94

Appendix A Systematic Review of the Literature 107

Appendix B Truce Related Media Statements in El Salvador 111

Appendix C August Town Five Year Peace Agreement 121

Appendix D Time Line El Salvador Gang Truce 123

Appendix E Social Violence Prevention Programming 124

i

EXECUTIVE SUMMARY

Background

While there is much literature describing the assumptions issues and effectiveness of crime

suppression (eg Decker 2003 Decker and Reed 2002 Katz and Webb 2006 McCorkle and

Miethe 2002) and prevention strategies (Esbensen and Osgood 1997) much less attention has

been paid to gang intervention programming particularly gang truces Little is known about

how often gang truces occur what conditions give rise to them the role of third parties in

brokering them their transformative effects and their effectiveness In this policy brief

sponsored by SolucionES1 and conducted by FUNDE a member of the SolucionES Alliance with

Arizona State University we systematically evaluate gang truces including reviewing prior

research and presenting evidence on the effectiveness of gang truces that have been

implemented in El Salvador Honduras and Jamaica for the purpose of identifying lessons

learned should other governments or donors wish to support gang truces in these or other

countries

Summary of Findings

We found that the truce in El Salvador resulted in a reduction in homicides that was not the

product of other trends or temporal factors Thus the truce in this country had a short-term

effect in reducing the lethality of violent crime though not necessarily in other dimensions of

insecurity By contrast the truces in Jamaica and Honduras resulted in no impact on violence

The Jamaican and Honduran experiences therefore mirror the results of prior gang truces that

1 SolucionES is multifaceted violence prevention program being implemented by an Alliance of five leading Salvadoran non-profit organizations who have come together to prevent crime and violence The Alliance members are Fundacioacuten Nacional para el Desarrollo (FUNDE) Fundacioacuten Salvadorentildea para la Salud y el Desarrollo Humano (FUSAL) Fundacioacuten Crisaacutelida (known locally as Glasswing) Fundacioacuten Salvadorentildea para el Desarrollo Econoacutemico y Social (FUSADES) and Fundacioacuten Empresarial para el Desarrollo Educativo (FEPADE) with partial funding from USAIDEl Salvador Together these five organizations have widely-recognized expertise in education health community development economic development research and youth leadership they are bringing their combined synergy and strengths to prevent crime and violence in El Salvador SolucionES is using a three-pronged strategy to prevent crime and violence in El Salvador 1) Strengthen municipal crime and violence prevention capabilities and actions 2) Increase social investment by the private sector to prevent crime and violence and 3) Research publish and disseminate findings to inform decision-making on crime and violence prevention This is one of 10 planned policy-oriented studies The full study is available on request

ii

have been studied to various extents including those in Los Angeles and Trinidad and Tobago

Important differences in how the various truces were negotiated may explain the different

results and one important difference ndash the ability of government and non-gang community

stakeholders to promise and immediately produce measurable deliverables ndash appear to be

especially significant

Prior Evaluations of Gang Truces

Little research to date has examined the effectiveness of gang truces and much of the research

that does exist has been restricted to the field of public health Studies of a gang truce in South

Central Los Angeles between the Crips and Bloods by Cotton (1992) and Ordog et al (1993

1995) found temporary reductions in the number of homicides and gunshot wounds during the

truce but these studies failed to point out that despite about a 35 percent decrease in

homicides for the first three months of the truce

homicides then doubled in months four through

eleven compared with the pre-truce period

Similar findings were reported in Trinidad and

Tobago (Maguire Katz and Wilson 2013) where

violence declined for a brief period of time

(again for about three months) but then

increased substantially over the long term (12

months) These results suggest that gang truces

may produce short-term benefits yet result in

long-term adverse consequences

Additionally these studies have suffered from a

variety of flaws including 1) poor theoretical

assumptions about gangs including often

ignoring that they might have the organizational

structure and internal discipline to enforce their

truces 2) relatively weak evaluation designs that fail to account for other potential causes for

What is a Gang Truce

A gang truce is a nonviolent resolution to a larger conflict between groups that has an impact on general levels of violence and other forms of criminality within a community (Ordog et al 1993 1995 Whitehill et al 2012) It differs from conflict interruption resolution or mediation efforts which seek to rapidly intervene typically through outreach workers or violence interruption specialists in episodic violent events between groups in a community Gang truces often involve dialogue and negotiations between multiple parties (eg gangs government NGOrsquos religious organizations) that seek to recalibrate the norms of conflict within and between groups for the purpose of reducing or eliminating violence and other crime

iii

or broader trends related to violence reductions 3) a failure to examine the processes involved

in creating gang truces 4) a failure to examine unintended consequences of the truces beyond

their impact on gun violence and 5) a lack of sophisticated statistical analysis capable of

discerning overall trends cyclic patterns outliers and turning points

Overview of Study Design

This policy brief examines gang truces implemented in El Salvador Jamaica and Honduras

including the processes undertaken with and between gangs and other stakeholders We

collected information about the processes associated with each gang truce from a variety of

sources including peer reviewed articles books reports local newspaper articles and

interviews of persons with first-hand knowledge about the gang truce in each nation We

examined the impact of each truce using official data We first performed a simple t-test

comparing the homicide rates before and after the truce We then employed time series

models not used in other studies to evaluate homicide rates as a function of time with truce

period indicators included to measure the effect of the truce net of the temporal trends We

also employed supplemental models to examine and control for factors other than the truce

that might have affected homicides over the study period

Research Findings

Implementing a Gang Truce

The gang truces studied have a number of common characteristics First in each case a

community was experiencing an uncharacteristically high number of gang related homicides

over a fairly lengthy period of time which resulted in each community placing strong pressure

on the government in general and the justice system in particular to respond to the problem

quickly and effectively Second each community had first attempted unsuccessfully to control

gang violence through suppression-oriented strategies Third each communityrsquos inability to

exercise traditional informal and formal social control to decrease levels of violence became

self-evident to the public and government This resulted in all involved stakeholders wanting to

seek (or participate in) an alternative strategy in which brokers would formally andor

informally work directly with gang leaders to establish a truce that would reduce homicides

iv

Key stakeholders involved in the negotiation and establishment of each gang truce were also

somewhat similar In each case examined the leaders of the largest and most violently

involved gangs were willing to participate In each of these cases it was clear that the gangs

sought to collaborate with the brokers not only for the purpose of reducing violence but

perhaps more importantly as a means to gain greater more positive recognition in the

community and to reap some form of benefit to themselves their members and perhaps their

community In each case while not always formally involved government officials were at a

minimum made aware of negotiations and in some cases solicited the assistance of third

parties to broker an agreement between stakeholders In each case it was at least implicitly

understood that the government would ldquolistenrdquo to the gang leaderrsquos expectations and offers

We found that when the government was no longer willing to ldquolistenrdquo to or collaborate with

brokers the truce processes ended abruptly Brokers were typically comprised of a very small

group (ie 2-3) of individuals who were perceived to be ldquohonest brokersrdquo In El Salvador and

Honduras this included a high ranking Catholic Church official a leader from an international

diplomatic organization (ie OAS) and other neutral parties In Jamaica this included a quasi-

governmental organization that had been established for the purpose of brokering negotiations

between gangs for the purpose of reducing violence as well as the local university which had

access to staff who were perceived to be neutral but had an interest in reducing violence near

the university

The strategies used to execute each gang truce were generally similar but had important

differences Each involved a team of brokers working to identify common goals to be achieved

and tangibles that could be delivered to the gang in exchange for the gang achieving the stated

goals They differed by how each partyrsquos promise was delivered In Honduras and Jamaica gang

leaders traded violence reduction pledges for long-term government and social changes such

as the development of substantial public works programs to reduce unemployment In

Honduras and Jamaica the government was asked to develop and deploy large scale social

programming in a short time framemdashsomething for which neither country had a strong track

record In El Salvador by contrast brokers secured promises for immediate changes in gang

behavior in exchange for feasible immediate deliverables from the government For example in

v

exchange for a reduction in gang violence the government agreed to immediately relocate

imprisoned gang leaders to less restrictive prisons and provide them some privileges Following

the successful execution of the first part of the Salvadoran truce which resulted in near term

success for both parties they began to negotiate broader issues that would take longer for both

sides to deliver Our findings suggest that some promised deliverables need to be easily and

quickly delivered by both parties early in the process to achieve trust and serve as a first test

of gang leadersrsquo ability to deliver Stakeholders have only a brief period of time to provide

promised benefits before trust is lost meaning that tangible benefits need to be delivered in

weeks or months not years

The Impact of a Gang Truce

El Salvador We found that El

Salvadorrsquos gang truce had a definite

impact on the homicide rate The

mean number of monthly homicides

declined from about 354 prior to the

truce to about 218 following the

truce for a net decrease of about

136 homicides per month Our data

show that between March 2012 and

June 2014 the truce had saved

about 5501 lives (see exhibit 1)

From a hypothetical stand it is

possible to make the assumption that a number of these deaths averted could have been

transformed in disappearances and therefore they were not counted within the official

homicide statistics However the results from the analysis point out that the number of

disappearances was not significantly related with the change in the global behavior of

homicides Additionally over the period of analysis (January 2010 to June 2014) there was no

significant change between the pre-truce and post-truce periods in the number of thefts

extortions robberies rapes and auto theftsrobberies

Exhibit 1 Forecast of homicides without gang truce

vi

We also found that the gang truce did not result in a homogenous decline in violence across

municipalities About 61 percent of municipalities experienced a decline in homicides but the

decline in violence varied substantially between municipalities We studied this issue further by

examining the impact of the initiative ldquoFree Violence Municipalitiesrdquo and it was found that the

behavior of violence in those municipalities was not significantly linked to the initiative but

rather to the general dynamic of the truce process nationwide Additionally we parsed out the

relative influence of the number of MS13 and 18th Street gang members on the street and in

prison from each municipality Our analyses indicated that following the truce the number of

MS13 and 18th Street gang members on the street in a municipality was not significantly

related to a decline in homicide but the number of imprisoned MS13 and 18th Street gang

members from the municipality was In particular the number of imprisoned MS13 gang

members from a municipality was associated with a significant decline in homicides in that

municipality following the gang truce and the number of imprisoned 18th Street members

from a municipality was associated with a significant increase in homicides in that

municipality following the truce These findings lend support to the idea that MS13 is more

organized than the typical street gang and that imprisoned MS13 gang members exhibit strong

influence over their fellow gang members on the street Our findings also suggest however

that the gang truce had a boomerang effect in municipalities with high numbers of imprisoned

18th Street members implying that 18th Street might not have as much organizational capacity

to regulate violence on the streets as MS13 The truce provided incarcerated MS13 and 18th

Street gang leaders an opportunity to negotiate with high-ranking officials and influential

diplomats including representatives of the Organization of American States This may have

increased their legitimacy inside and outside of their gangs It appears that MS13 was able to

exert its span of control over the communities in which they had influence and they were able

to deliver on the terms of the gang truce negotiations In the case of 18th Street however

incarcerated gang members may not have had the same organizational capacity for

communicating and carrying out directives In fact a review of the gang truce indicated that

there was a conflict taking place between two factions within 18th Street Consequently the

organizational structure and culture of 18th Street might be more diffuse than that of MS13

vii

and its leadership structure might not be as strong because of the internal fractures within the

gang This might further explain why homicides increased in 18th Street communities The

internal fractures within the 18th Street gang may have resulted in intra-gang violent conflict

that was largely contained within 18th Street controlled territories

Jamaica In Jamaica at first glance our impact findings appeared to show that the gang truce

might be an effective mechanism for reducing violence Bivariate analyses showed a significant

decline in homicides immediately after the truce was implemented This explains the work

previously published by policymakers researchers and news reporters Upon closer

examination of the data however comparing change in the target area to other areas in

Jamaica and accounting for temporal trends we found that the decline in homicide was part of

a larger nationwide decline in violence and that the gang truce was not responsible for the

decline The only significant effect that we uncovered was the possibility that homicides were

displaced outside the target area for a brief period of time but then returned to previous

levels

Honduras Our impact findings from our analysis of data from Honduras tell a very similar story

as Jamaica Initial analysis showed that the number of homicides on average declined across

municipalities following the gang truce Specifically the mean number of homicides declined by

12 per 100000 population from an average of 687 per 100000 population in each

municipality before the truce to an average of 566 thereafter However after we examined the

effect of the truce through time series analysis and included a variable (month) to control for

the temporal trends in the data the impact of the truce we observed in our bivariate analysis

was no longer significant Our findings as in Jamaica suggest that the decline in homicides was

less a consequence of the gang truce than of a broader short to medium term trend And

importantly in both Jamaica and Honduras the respective governments failed to deliver on

gangsrsquo demands for large-scale social and employment programs

Conclusions and Policy Implications

Over the last several years there have been a number of naturally occurring experiments

involving gang truces in a variety of nations and regions Findings from evaluations of gang

viii

truces are mixed In El Salvador the gang truce could be characterized as an effective short

term strategy to reducing homicides It is worth mentioning that despite homicides rates are

above truce levels they continue slightly below pre-truce levels In Jamaica and Honduras the

gang truce had no short or medium-term impact on overall violence In Los Angeles and

Trinidad y Tobago there was evidence that violence decreased for at least ninety days but then

increased substantially beyond those rates observed prior to the gang truce Thus the

effectiveness of the truce in El Salvador appears to be isolated and must be evaluated within

the context of other truces that have failed to reduce homicide violence Policy makers must

evaluate whether the conditions that allowed short term effectiveness of the gang truce in El

Salvador (such as the ability to promise and deliver immediate results) exist in other violent

areas before evaluating whether a truce strategy might be appropriate And they should be

heavily cautioned that the potential for long term negative consequences might outweigh the

potential for short term benefits

Indeed it is important to note that a number of scholars have noted that gang truces are likely

to result in a boomerang effect with gang violence increasing over the long run because of

enhanced cohesion within the gang (Klein 1995) Maguire (2013) notes that when government

officials negotiate a truce with gangs they might ldquoinadvertently be acknowledging gangs as

legitimate social entitiesrdquo (p 11) This in itself might increase cohesion among gangs which has

been found to be associated with increased levels of criminality (Decker et al 2008 Klein 1971

Maguire 2013) Hence it is important to consider the fact that gangs are illegal groups in El

Salvador and it should be cautious when carrying out dialog or negotiation processes with

them Further research is needed to examine how gang truces might impact group cohesion

and if it does whether the cohesion created could be effectively directed toward more

productive non-violent endeavors Gang truces convey the well-intentioned image that violence

has been addressed and policymakers are doing something about the problem but unless the

truce is implemented in a manner and under conditions where immediately achievable results

can be promised delivered and measured there remains a significant chance that the truce will

fail or worse yet backfire Thus it is imperative that any type of concession made by

ix

Governments to gangs within a truce framework should be transparent so that all sectors of

society have certainty that every action is being done within the existing rule of law2

In the case of El Salvador the truce arises from to the absence of effective public policies and

practices for violence control and prevention The truce was planned as a strategy to reduce

gang-related homicides During the process different organizations got involved including

religious (facilitating and protecting human rights) non-government (managing and facilitating

dialog and negotiation processes) international (providing funding for insertion programs) and

government (facilitating and providing certain conditions for dialog and negotiation)

organizations Some of the concessions that the Government provided in order to achieve a

reduction of homicides were within the law but others generated confusion and they seemed

to be close to the legal or socially acceptable limits This fact along with the poor transparency

of authorities in the management of the practice with public media and public weakened the

process and postponed its continuity

The present study suggests that gang truces should only be used as a means of last resort

and then only under certain conditions Given the risks associated with a gang truce

communities with high levels or at least modest levels of formal social control should rely on

other more promising gang control strategies such as pulling levers (ie Boston Ceasefire)

community oriented policing and the Gang Resistance Education and Training (aka GREAT)

program Only when the state has limited or greatly reduced capacity for social control should a

truce be considered Concomitantly a gang truce should be considered as an alternative only

when a community is experiencing a substantial amount of gang violence Communities that

are experiencing minimal to modest amount of gang violence may risk more from the

establishment of a gang truce than they have to gain Additionally our findings suggest that a

gang truce might only be feasible when gangs are sufficiently well organized to be able to

regulate their membersrsquo behavior and cause their members to behave less violently In El

22 Pragmatically this issue is complicated On the one hand transparency is a foundational element within a democracy and is necessary to ensure proper oversight of the government On the other hand it might not be possible to implement a gang truce with too many actors having a voice Policymakers might consider creating a policy that allows such negotiations take place but requires particular actors (such as a judicial body) to be informed of the process to ensure transparency and adherence to the rule of law

x

Salvador there is evidence of the strong organizational structure of gang MS13 Among other

factors the magnitude of its membership the chain of command from its leaders in prison and

the discipline of its leaders in the streets seemed confirmed On the contrary the

organizational structure of gang Barrio 18 ndash divided in two factions fighting over the leadership

ndash showed to be a less stable counterpart within the truce

Finally dialog and negotiations processes with or between gangs must have the capacity to

promise and deliver immediate benefits to the gangs that gang members can see or experience

in order to secure their continued participation in the truce as well as the capacity to monitor

and respond to truce violations Most importantly any effort aimed at reducing violence is

important and should be examined and assessed but it must have a transparent foundation

especially when it affects population rights as a whole

General recommendations

Gang truces are conjunctural strategies States who suffer from gang-related violence must

establish permanent public policies for crime control and prevention A government that

considers implementing a gang truce should be aware that it cannot become the center

strategy of its public policy for citizen safety

Gang truces should only be used as a means of last resort and then only under certain

conditions Stakeholders must determine whether a process of dialog or negotiation with gangs

is legal ethical and feasible

Stakeholders must anticipate demands that are likely to arise and their response options Some

demands may be easily met such as improved prison conditions Others are much more

difficult and amorphous such as community development through more integrated violence

prevention programs (such as those implemented by SolucionES in El Salvador) local economic

development programs or economic reinsertion of ex-gang members

Stakeholders should incorporate immediately achievable and demonstrable deliverables Long-

term goals and promises are unlikely to create the trust needed to sustain a gang truce

Stakeholders must first determine the position in which they are negotiating the incentives that

are possible to deliver and the boundaries and limits they face Gangs are mostly likely to trust

xi

representatives from NGOs community-based organizations and members of the faith based

community as brokers because they are considered more reliably neutral advocates for peace

They need to understand the capacity of the government to deliver promises in a timely

manner

Governments have to make a choice about the visibility and transparency of its participation

This decision needs to be made in the context of the national and local laws the publicrsquos

expectations of transparency and patterns of practices of the past

Governments must be strategic in their support for truce initiatives Some donor funded

programs run by the government prohibit gang member participation and if the government

does not receive approval from the donor it may risk the donor withdrawing its sponsorship of

the program

Governments must ensure an inter-institutional coordination for the management of truces to

avoid the responsibility to be of a single government institution It is necessary to generate or

collect reliable and pertinent data that can be used to analyze and assess the process

It is necessary to implement an effective monitoring system of the truce process similar to that

used in the full report as it can help parties understand what is working and who is delivering

on their promises More specifically the monitoring and truce management system should be

able to identify truce violations and be prepared to respond through the use of legal and

effective practices if stakeholders do not comply

Finally it is necessary to develop evaluations of gang truces and monitoring programs and

support violence prevention activities local economic development activities and pilot

programs to support the reinsertion of ex-gang members into society Clearly national

governments municipal governments NGOs and community-based organizations need

increased capacity and resources to discourage the growth of gangs among at-risk youth It is

therefore increasingly important to create economic opportunities for gang members willing to

leave the gangs and find other legal employment Developing and sustaining those

opportunities in nations with high incidences of poverty will require significant international

funding

xii

References

Cotton Paul 1992 ldquoViolence Decreases with Gang Trucerdquo JAMA 268(4) 443-44

Decker Scott H 2003 Policing Gangs and Youth Violence Belmont CA Wadsworth

Decker Scott H and W Reed 2002 Responding to Gangs Evaluation and Research

Washington DC National Institute of Justice

Decker Scott H Charles M Katz and Vincent J Webb 2008 ldquoUnderstanding the Black Box of

Gang Organization Implications for Involvement in Violent Crime Drug Sales and

Violent Victimizationrdquo Crime and Delinquency 54 153-72

Esbensen Finn-Aage and D W Osgood 1997 ldquoNational Evaluation of GREATrdquo US

Department of Justice Office of Justice Programs National Institute of Justice

Katz Charles M and Webb Vincent J 2006 Policing Gangs in America New York Cambridge

University Press

Klein Malcolm W 1995 The American Street Gang New York Oxford University Press

mdash 1971 Street Gangs and Street Workers Englewood Cliffs NJ Prentice-Hall

Maguire Edward Charles Katz and David Wilson 2013 ldquoThe Effects of a Gang Truce on Gang

Violencerdquo Unpublished paper Washington DC American University

Maguire Edward 2013 Research Theory and Speculation on Gang Truces Woodrow Wilson

International Center for Scholars in Washington DC

McCorkle R and T Miethe 2002 Panic The Social Construction of the Street Gang Problem

Upper Saddle River NJ Prentice-Hall

Ordog Gary J W Shoemaker Jonathan Wasserberger and Michael Bishop 1995 ldquoGunshot

Wounds Seen at a County Hospital Before and After a Riot and Gang Truce Part Twordquo

Journal of Trauma-Injury Infection amp Critical Care 38(3) 417-19

Ordog Gary J Jonathan Wasserberger Julius Ibanez Michael Bishop Eduardo Velayos

Subramaniam Balasubramanium and William Shoemaker 1993 ldquoIncidence of Gunshot

Wounds at a County Hospital Following the Los Angeles Riot and a Gang Trucerdquo Journal

of Trauma 34 779-82

Whitehill J M Webster D W amp Vernick J S 2012 ldquoStreet conflict mediation to prevent

youth violence Conflict characteristics and outcomesrdquo Injury prevention injuryprev-

2012

1

Introduction

Given the increasingly devastating individual family and community effects of gang activity

over the past several decades an increasing body of literature has focused on gangs gang

members and gang activity A core theme running throughout this body of literature is that

gang members are significantly more likely to be the offenders and victims of violent crime than

non-gang members (Battin Hill Abbott Catalano and Hawkins 1998 Esbensen Winfree He

and Taylor 2001 Katz et al 2011 Huff 1998 Curry et al 2002 Miller and Brunson 2000

Miller and Decker 2001 Pyrooz et al 2012 Curry et al 2001) and disproportionately affect

neighborhood levels of crime and violence (Block 2000) These findings have been robust in

that they have been repeatedly found regardless of research methodology (ie use of official

data self-report data observational data) (eg Curry 2000 Decker 1996 Deschenes and

Esbensen 1999 Esbensen et al 2001 Gordon et al 2004 Katz Webb Schaefer 2000 Webb

Katz Decker 2006) or research setting (ie North America Europe Asia South America

Australia) (Klein and Maxson 2006) As a consequence it should not be surprising that

policymakers academics and community leaders have focused much of their attention on

developing responses to address community gang problems

Suppression strategies have been the favored public policy response to gangs since the 1980s

(Venkatesh 1999 Spergel et al 1995) Suppression strategies typically rely on focusing

criminal justice resources on gang members through such practices as targeted and enhanced

police patrols intelligence databases vertical prosecution and enhanced sentences for those

convicted (Katz and Webb 2006) Suppression strategies are based on deterrence theory and

are founded on the principal that swift certain and severe penalties for those involved in gangs

and gang activities will necessarily result in fewer individuals joining gangs and will deter people

from engaging in gang crime (Klein 1995) By the early-to-mid 1990s as gang problems

continued to proliferate and policymakers sought alternative gang control strategies gang

prevention programming flourished (Papachristos 2013) Gang prevention programs were

aimed at the general youth population or focused on at-risk youth or neighborhoods Gang

prevention programming was based on the premise that by reducing risk factors and increasing

2

protective factors prevention specialists could inoculate youth from gang membership

(Esbensen 2000) Prevention programming has most frequently come in the form of school-

based prevention programming such as the Gang Resistance Education and Training (GREAT)

program where students are exposed to a curriculum designed to reduce both participation in

gangs and gang crime as well as the SolucionES3 program which engages youth and their

families through integrated community-based and school based programs

While a burgeoning body of literature has developed describing the assumptions issues and

effectiveness of suppression (eg Decker 2003 Decker and Reed 2002 Katz and Webb 2006

McCorkle and Miethe 2002) and prevention strategies (Esbensen and Osgood 1997) much less

attention has been given to examining gang intervention programming This might largely be

the consequence of the absence of such programming over the past 30 to 40 years By the

1980s policymakers no longer believed that social intervention approaches were an effective

strategy to control gangs and gang problems Although gang intervention strategies took many

forms they were based on the assumption that gang membership is the by-product of a socially

deprived community and that the values and norms of gang youth can be influenced and

directed toward those of mainstream society As such intervention programs often focused on

diverting youth from gangs or sought to minimize the consequences of gangs and gang activity

Gang intervention strategies include crisis intervention dispute resolution street-level

counseling and youth outreach (Spergel 1995) However many policymakers community

organizations and academics have argued that such approaches not only did not reduce gang

activity but may have lead to increased group cohesiveness which in turn may have lead to

increased delinquency (Klein 1971 Spergel 1995) More recent research has yielded similar

results For example a number of studies examining replications of Chicago CeaseFireCure

3 SolucionES is a Global Development Alliance in which $20 million is provided by USAID and SolucionES Alliance

members will contribute an additional $22 million dollars during a 66-month period starting July 2012 SolucionES is being implemented by an Alliance of five leading Salvadoran non-profit organizations who have come together to prevent crime and violence The Alliance members are Fundacioacuten Nacional para el Desarrollo (FUNDE) Fundacioacuten Salvadorentildea para la Salud y el Desarrollo Humano (FUSAL) Fundacioacuten Crisaacutelida (known locally as Glasswing) Fundacioacuten Salvadorentildea para el Desarrollo Econoacutemico y Social (FUSADES) and Fundacioacuten Empresarial para el Desarrollo Educativo (FEPADE) Together these five organizations have notable expertise in education health community development economic development research and youth leadership they are bringing their combined synergy and strengths to prevent crime and violence in El Salvador More information on social violence prevention programming is presented in Annex E including a summary of the violence prevention initiatives being implemented by SolucionES

3

Violence which relies heavily on crisis intervention dispute resolution street-level counseling

and youth outreach have found these strategies to either be ineffective or worse increase

levels of violence (Fox et al 2014)

One type of gang intervention gang truces however has received little attention in the

literature This is somewhat surprising given its use and public claims of effectiveness The

purpose of this report is to systematically examine the effectiveness of gang truces Gang truces

have largely been a ldquoblack boxrdquo that is little is known about the frequency of their use

conditions that give rise to them the role of third parties in brokering gang truces the

transformative effects of truces and the effectiveness of gang truces In this report we review

prior research on gang truces and present evidence on the effectiveness of gang truces that

have been implemented in El Salvador Jamaica and Honduras

Theoretical and Policy Rationales for Gang Truces

Much of the concern about gangs over the past twenty years has been their close association

with violence Qualitative and quantitative research has repeatedly found that gangs and gang

members are involved in high levels of serious violent offending Decker (1996) attributes the

gang-violence relationship in part to the collective and normative structure of gangs which

supports and encourages the use of violence both preemptively and in retaliation He further

explains that gang membership encourages the use of violence in retaliation against threats

and attacks whether actual or perceived which results in patterns of inter-gang conflict

characterized by escalating violence Related Cooney (1998) points out that gangs are not all

that different from ldquowarrior societiesrdquo He argues that while gangs are violent the violence

that they engage in is different than that engaged in by non-gang members in that it commonly

takes on a feud-like dimension A perceived slight violation of turf or other disrespectful

action might invoke a shooting which in turn evokes a retaliatory shooting which in turn

results in another retaliatory shooting creating a self-perpetuating cycle of violence

Early research and theoretical work examining gangs and gang conflict suggested that much

violent gang behavior was the function of status management (Thrasher 1927 Whyte 1943

Short and Strodtbeck 1965) Borrowing from the sociology and social psychology literature on

4

impression management theory gang scholars hypothesized that youth place significant

emphasis on image managementmdashseeking to impress their peers and limiting the potential to

embarrass themselves in front of others (Hughes 2005) These early gang theorists postulated

that violence is an instrument used by gangs and gang members to achieve manage and

protect status They contend that gang members often make decisions to become involved in a

violent conflict based on rational processes that weigh the immediate loss or gain of status

within the gang against the relatively small probability of being formally sanctioned by officials

within the criminal justice system (Hughes 2005) A number of researchers have examined the

relationship between status considerations and gang violence and have found the association

to be particularly robust regardless of gender (Campbell 1991 Giordano 1978) ethnicity

(Anderson 1990 1998 Vigil 1988) and location (Jankowski 1991 Spergel 1995) Hughes

(2005) notes that the centrality of status consciousness by gang youth may account for the

existence of the facilitative gang effect that gang scholars have repeatedly observed (ie Gangs

increase the amount of delinquency individuals are involved in beyond that of a group of

delinquent peers (Gordon et al 2004 Thornberry et al 2003)

Another micro-social factor associated with gang violence is group cohesiveness While our

understanding of the relationship between gang cohesiveness and violence is underdeveloped

some scholars have reported a strong relationship between the two In particular Klein and his

associates (1971 1967) and Lucore (1975) have reported that increases in gang cohesiveness

lengthens periods of gang membership and increases members participation in gang crime and

violence Cooney points out that there are strong relational ties between gang members that

necessarily result in increased cohesiveness among members For example gangs are

comprised of neighborhood youth who share common cultural and economic experiences and

have often known one another for lengthy periods of time He also points out that gangs are

groups that have strong self-proclaimed and formalized identities (eg colors symbols names

monikers) and have at least some organizational structure Their sense of group is maintained

by their common understanding of their members and friends and their attachment to their

territory (or neighborhood) Decker (1996) notes that the relatively high level of group

5

cohesiveness exhibited by gangs facilitates both collective behavior and liability (For exception

see Hughes 2013)

Accordingly both micro-social factors (ie status management gang cohesiveness) serve to

augment levels of gang violence and make it difficult for third parties to intervene Violence

within the context of gangs serves as a form of informal social control Gangs and gang

members cannot seek assistance from legitimate institutions of social control to solve conflicts

because they would risk losing status (Anderson 1999) Likewise the collective nature of the

gang not only increases potential offenders and targets of violence but also facilitates at the

group level the need for retributive justice on the behalf of injured members Moreover gang

members desire to impress others with their commitment to the group and use violence to

demonstrate their commitment to their group and to increase their status within the group All

of this results in an increased cycle of gang conflict and violence4

Over the past several decades residents and policymakers have expected the police to address

violent gang conflicts However there are several limitations to the police response to gangs

First as noted above many of the same factors that are associated with violent gang conflict

also limit the effectiveness of the police to have an impact on violent gang conflict Gang

members are not going to contact the police to resolve a conflict because it could result in loss

of status and expose them to the police discovering illegal activities that they are involved in

(Katz 2001 Katz and Webb 2006) Related citizens in neighborhoods with gang problems are

also reluctant to call the police out of fear of gang reprisals (Katz 1997 Webb and Katz 2003)

or because they have a poor perception of the police themselves (Katz Choate and Webb

2002) Third the police response to gangs in most communities is a reactive response to a

specific incident after it has occurred rather than a proactive response to intervene in ongoing

disputes between gangs (Katz and Webb 2003 2006) Most police agencies simply do not have

4 There is anecdotal evidence for these hypotheses Hughes and Short (2005) through field observations and interviews with members of 20 Chicago gangs examined the micro-social context of violent-related incidents Specifically they were interested in the specific social conditions that precipitated violence Their analysis revealed that gang disputes associated with retaliation escalated into violence significantly more often (ie 55 of the time) than disputes associated with normative order violations or an identity attacks which only resulted in violence roughly 33 percent of time

6

the intelligence networks required to intervene in gang conflicts until after they have risen to

relatively high levels (Katz 2003) Fourth police suppression strategies have been linked to

increasing gang cohesiveness and possible increases in gang crime (Klein 1995)

Some policymakers and community activists have proffered that an alternative to reliance on

formal mechanisms of social control such as the police is the gang truce (Spergel 1995) The

goal of a gang truce is to reduce or even eliminate violent conflict between those gangs that are

warring with one another As such compared with other strategies that often seek to reduce

general levels of crime a gang truce has the very specific goal of reducing violence between

two or more gangs that are in conflict with one another Unfortunately the literature provides

little guidance on the theoretical assumptions of why a gang truce should reduce inter-gang

violence Henderson and Leng (1999) hypothesize that at the root of gang truces is the notion

that they involve the renegotiation of existing norms within and between gangs The authors

claim that as a violent dispute escalates between gangs leaders and members are placed in the

situation of appearing weak to both members of their own gang and to members of the rival

gang if they do not respond with the appropriate amount of force or if they were to suggest a

peaceful resolution to the dispute Accordingly Henderson and Leng argue that as the cycle of

violence escalates between two or more gangs behavioral norms shift toward the increased

valuation of violence to resolve the conflict because it is the only option readily available to

them A gang truce on the other hand which is often mediated by a third-party is believed to

break the cycle of violence by providing the gangs involved in the dispute with a cooling-off

period (Spergel 1995) In the interim period new norms of expected behavior within and

between gangs are established In other words the cooling-off period is believed to recalibrate

norms of behavior that are more consistent with the security interests of the gang and its

members (Henderson and Leng 1999)

For many of the reasons stated above a number of communities have participated in gang

truces Gang truces have been observed in the United States Central America and the

Caribbean and gang leaders in other nations are considering its implantation (Fahah 2012)

Unfortunately little is known about gang truces We do not know when they come into

consideration how they are implemented whether they decrease increase or even have a

7

significant impact on violence In the below section we systematically review the existing body

of literature on gang truces Appendix A describes the methodology we employed to conduct

the systematic review

Prior Research on Gang Truces

Little research to date has examined the effectiveness of gang truces and of the research that

has been conducted has been restricted to the field of public health The most celebrated

truce to date was Philadelphiarsquos House of UMOJA 1974 gang truce In preparation of the truce

gang leaders in prison were called upon for their support along with key community leaders

from churches businesses schools and the police When UMOJA called for the gang summit it

was reported that 500 gang members and 75 of Philadelphia gangs attended By the

conclusion of the gang summit a 60-day truce was announced which reportedly resulted in no

gang members being killed over the 60-day period However no systematic evaluation of the

gang truce was conducted (Woodson 1981)

The first attempt to evaluate a gang truce to our knowledge was conducted by Cotton (1992)

who examined the results of a gang truce in South Central Los Angeles between the Crips and

Bloods Data provided by the police department indicated that over the 6-week period when

the truce took place drive-by shootings decreased by 48 compared to the same 6-week

period the prior year decreasing from 162 to 85 Likewise gang-related homicides dropped by

62 from 26 to 105

Ordog et al (1993 1995) examined the effects of a gang truce in Los Angeles using emergency

room admissions data Specifically the authors examined changes in the daily and monthly

number of gun shoot wound (GSW) emergency room admissions before during and after a

gang truce The catchment area for the emergency room was 100 square miles The authors

noted that while they were able to clearly identify the date the gang truce began because of

the media coverage that it received there was no specified date that the truce ended and as a

5 The authors did not report whether the drop in gang-related homicides was citywide or in the South Central Los Angeles neighborhood where the gang truce took place

8

consequence after 12 weeks the authors regarded it as called off for evaluation purposes

because it was no longer being discussed in the media and gun shot wounds began to increase

Student t-tests were used to examine changes before during and after the gang truce Ordog

et al (1995) reported that there were approximately 7 GSW admissions per day the 12-months

preceding the truce compared to 45 GSW admissions per day during the gang truce and 126

GSW admissions per day in the 11-months following the gang truce The authors concluded

that their analysis ldquoclearly showed that the institution of a gang truce had reduced the number

of GSW victims seen in an hellip inner city Level I trauma centerrdquo (p 419)6

However it is important to point out that while the gang truce in Los Angeles did decrease

homicides by about 35 percent for the first three months it then doubled in months four

through eleven compared with the pre-truce period - a pattern that the authors did not

discuss Similar findings were reported in Trinidad and Tobago (Maguire Katz and Wilson

2013) where it was determined that homicides declined for a brief period of time (again for

about three months) but then increased substantially over the long term (12 months) These

results suggest that gang truces may produce short-term benefits yet result in long-term

adverse consequences

While the research examining gang truces shows their potential promise and their potential for

greater harm we believe that the findings should be viewed with caution for two reasons

First some of the implicit theoretical assumptions that gang truces are built on may not be

accurate Prior research on gangs suggests that they have limited organizational structure

(Decker Katz and Webb 2008 Decker Bynum Weisel 1998) and have few formal

mechanisms to influence member behavior For example most gangs do not have formal

leaders do not require members to pay dues and members do not make contributions to the

gang for the purpose of developing the gangs infrastructure (ie guns housing etc) A number

of academics have pointed out that ldquosophisticated gang organizations is still largely a product of

the self- or organizational-interested musings of gang leaders certain police officials academic

6 Similar results were presented by Ordog et al 1993 using the same data but analyzing it over a shorter period of time and aggregating the data by month instead of by day

9

researchers media reporters based on very limited hard datardquo (Spergel 1995 79-80)

Therefore even if a truce was successfully negotiated between members of gangs in conflict

much of the academic research suggests that gang leaders do not have enough control over

members to enforce a truce Related prior research on gangs and gang members suggests that

they are primarily comprised of young people with few ldquodiplomaticrdquo skills (Henderson and Leng

1999) The very same organizational and normative features of gangs that result in gang

violence (ie often bastions of young minority male street youth) are those same features that

most likely limit the likelihood of a gang having the capacity to abide by a gang truce In sum

implicit in the theoretical assumptions of a gang truce is that gangs have the organizational and

cultural capacity to create and maintain (at least for a short while) a truce which is inconsistent

with much prior research on the organizational characteristics of gangs

The second reason we believe that prior evaluations of gang truces should be viewed with

skepticism is that they have relied on relatively weak evaluation designs For example prior

studies examining gang truces have not incorporated the use of comparison areas or control

groups The causes of reductions in gang violence found in previous evaluations might be

many For example the Los Angeles riots took place just before the gang truce that Ordog et al

evaluated Zinzun (1997) reported that gang culture and violence changed briefly but abruptly

following the riots because gangs and gang members in part redirected their anger and focus

toward the police As such the decline in GSW admissions may have been the consequence of

an overall city-wide decline in gang violence in the wake of the LA riots Related prior research

has relied on fairly broadly defined outcome measures such as GSW admissions and general

levels of drive-by shootings Such outcome measures lack specificity in terms of attributing

gang violence to the specific gangs involved in the gang truce Evaluation designs used in

previous studies have also lacked specificity in terms of the catchment area where violent

activity was labeled as gang related For example Ordog et alrsquos (1993 1995) outcome measure

included all emergency room admissions for GSW However the emergency room received

patients from a 100 mile square area surrounding the hospital an area that was most likely

much larger than the gangsrsquo territories involved in the truce

10

Additionally prior research examining gang truces has not examined the processes involved in

the creation of the gang truce No context has been provided in terms of the factors that lead

to the gang truces whether the gangs were pushed or pulled into truces whether outside

parties helped mediate the truces or whether on-going mediation was required to maintain the

truce In other words we still do not know the processes related to the formation of a

successful gang truce Instead prior research and evaluation has treated the gang truce much

like a black box where it is described in very general terms but its details are not revealed

Likewise prior evaluations have not examined the effects of a gang truce beyond its impact on

gun violence A number of prior researchers have argued that mediating such activities serves

to legitimize gang leaders increases gang identity for members and results in greater group

cohesion (Haskell and Yablonsky 1982 Klein 1995) As such we do not know if there are any

unintended consequences resulting from gang truces due to the methodologies chosen for

prior research and evaluation Finally prior research on gang truces has not relied on more

sophisticated statistical analysis that has the capacity to not only discern overall trends but also

cyclic patterns outliers and turning points

THE PRESENT STUDY

The present study seeks to understand the effectiveness of the gang truce when negotiations

occur with or between gangs and with government institutions or civil society organizations

We have briefly reviewed the theoretical and policy rationales for gang truces and the

literature and prior research on this form of gang intervention Next we present three case

studies one from El Salvador another from Jamaica and another from Honduras Each offers

evidence about their impact on violence and lessons learned that are unique to their particular

circumstances In conclusion the case studies are considered together and we present their

collective policy implications

11

Case Study A Gang Trucemdash The Salvadorian Experience

Over the past two years the Salvadoran gang truce brokered by an array of local policymakers

and international development organizations has attracted national and international

attention Violence in El Salvador has been at an all time high with a rate of 692 homicides per

100000 population in 2011 (UNODC 2014) making El Salvador one of the most violent nations

in the world (Parkinson 2014) Since 1992 government and non-government actors have been

responding to the rising tide of gang violence using traditional suppression strategies such as

increased policing legislative changes and more severe prison sentences These traditional

mechanisms of formal social control however were proving ineffective if not counter-

productive (Perez 2003)

In response to the inadequacy of traditional strategies stakeholders altered their course in an

effort to radically reduce gang violence in the nation Members of the Funes administration led

a group of negotiators comprised of the Catholic Church a former congressman and the

Organization of the American States (OAS) to help frame the conditions for a possible truce

between the MS13 and 18th Street gangs (Umana de Leon and Tager 2014) In March 2012 a

truce was reached The goal of the gang truce was to reduce violence specifically gang-involved

homicides Included in the terms in exchange for the gangs acting to reduce homicides certain

incarcerated gang members were to be transferred to lower security prisons to receive

sentence reductions and special visitation privileges and to be permitted more communication

with those outside of the prison for the purpose of conducting crisis interventions to mitigate

the violence (Salanegra 2012)7 The gang leaders also agreed to no longer recruit children into

their gangs reduce violence against women give up a small number of guns and continue to

participate in negotiations (Seelke 2014 11-12)

7 It is important to note that there were other discussions that took place during the mediations that included as a local publication noted ldquoProhibition Act Gang send the army to barracks end the police operations in controlled areas by gangs repeal of the figure of the witness criteriado (with criteria of opportunity or ldquowitness under a plea agreementrdquo) and a number of improvements in the quality of life of the inmates and its familiesrdquo Source httpwwwsalanegraelfaronetes201209cronicas9612

12

The present case study seeks to understand the role of negotiations with or between gangs and

government institutions or civil society organizations and their impact on violence in El

Salvador The specific objectives of this case study were to (1) Identify and document the

negotiation processes with or between gangs This includes but will not be limited to

identifying the actors involved in the negotiations the goal(s) of the negotiations and

strategies employed to carry out negotiations (2) Identify the impact of negotiations with

andor between gangs on violence and other forms of criminality and (3) Present conclusions

about the strengths and weaknesses of negotiations and their potential for positively or

negatively reducing crime in Salvadoran society

The Salvadorian Gang Truce

While the origins of the Salvadorian gang truce are somewhat unclear some accounts suggest

that in the fall 2011 the Minister of Security (then David Munguia Payes) mentioned the idea of

starting a dialogue between the gangs to Raul Mijango (a former congressman) Shortly

thereafter Mijango spoke to Monseigneur Fabio Colindres the military chaplin about the idea

They started to communicate with gang leaders in January 2012 From the beginning the

Minister of Security stated that he would not personally communicate with gang leaders and

noted that the official position of the government was not to negotiate with offenders But he

did allude to the fact that the gang truce was part of a new strategy to address the nationrsquos

gang problem (Archibold 2012) The Salvadorian gang truce was multi-dimensional involving a

varying number of actors communication styles and tactics In the section below we discuss

the parties involved in the truce the negotiation process and the strategies used by the gangs

and government in furtherance of the truce

Parties involved in the truce There were a series of persons groups and other entities who

directly or indirectly participated in the negotiation process whether they were negotiators or

collaborators From the beginning negotiations between the MS13 and 18th Street gangs were

considered confidential and were held in strict confidence between the gangs and the

negotiators Beginning in early 2012 the first two mediators to serve were a former

congressman (Rauacutel Mijango) and a priest (Fabio Colindres) These individuals were able to gain

13

the trust of the gangs and also had open communications with the government and

community

As the process moved forward however church leadership changed its initial position and

withdrew its support of the truce This left the priest to participate as a private individual no

longer serving as a representative of the church The mediators had some early success

however as the mediators changed the lines of communication became less clear and less

reliable Still they played an important role in the process Finding individuals whom the gangs

would trust and who could speak for government and community institutions was challenging

Further as government ministers and presidents changed it became less clear whether or not

the government was participating in the dialogue between various parties much of the

confusion in negotiations was related to understanding the governmentrsquos official role in the

process

In the beginning the negotiating parties included the two mediators and MS13 and 18th Street

the two predominant gangs in El Salvador MS13 is the largest gang in the country with about

250 cliques throughout the nation For the most part its first- and second-generation leaders

were in prison There are municipalities in which only MS13 operates and therefore its

members never clashed with 18th Street adversaries MS13 members are largely involved in

extortion violence and intimidation of the public The MS13 leadership has been shown to

have greater control of its organization than most its members have been more disciplined

and despite its size the gang seems to have sufficient and effective communication

mechanisms between cliques and its leadership (ldquoranflardquo or ldquoruedardquo)

The 18 Street gang is somewhat smaller than MS13 Long before the truce process began

factions existed within 18th street which had become deeply involved in internal conflicts over

leadership structure Those differences had become more pronounced over time leading to the

establishment of two movements known as the ldquoSurentildeosrdquo (Southerners) and the

ldquoRevolucionariosrdquo (Revolutionaries) that participated in the peace process but under different

14

representation Despite their differences however for the negotiation process the two

fractions of 18th street established similar demands between themselves and with MS138

As noted above the governmentrsquos involvement in the truce process was never clearly

articulated and it varied with ministerial changes The Funes Administration specifically the

Ministry of Justice and Public Security (MJPS) had been engaged in the negotiation process

early on In 2013 however with the change in MJPS leadership this also changed Tensions

grew over the role of the government in the truce negotiations According to the mediators the

new Minister of Security wanted to dismantle the process and for this reason he blocked the

mediatorsrsquo entry to the prisons where they had been meeting with gang leaders The change in

government leadership brought changes in strategy As with all negotiations mutual trust was

the cornerstone When the personnel and their agendas changed so did the trust in the peace

process

Late in 2012 as negotiations began to focus on the relationship between the gangs and the

community other organizations were called upon to help facilitate this interaction A technical

committee was created to oversee progress towards the violence reduction goal this was

announced at the end of August 2012 The committee moved forward with support from the

Organization of American States (OAS) It coordinated actions intended to improve the living

conditions of incarcerated persons to reintegrate into society those who had served their time

in prison and to prevent violence and provide assistance to victims of violence The committee

operated as a mechanism for coordinating the negotiation process with technical support from

the United Nations Development Program (UNDP)‒El Salvador Interviews with experts

indicated that Fundacioacuten Humanitaria with support from several organizations like Interpeace

and others had also supported the operations of the mediators and served as an important

point of reference for the gangs

In each territory where agreements were made government representatives from some

municipalities were invited to participate in the process as key implementation stakeholders

Between December 2012 and January 2013 the municipalities of Ilopango Santa Tecla

8 httpsalanegraelfaronetes201110cronicas5645

15

Sonsonate and Quezaltepeque became among the first to become part of what later became

known as the Sacred Municipalities (aka violence free municipalities) By November 2013

several other municipalities had become part of this group La Libertad Apopa San Vicente

Puerto El Triunfo Nueva Concepcioacuten Ciudad Delgado and Zacatecoluca

The gangs and the government offered several incentives to one another The gangs offered a

number of actions to reduce the incidence of crime (ie cease of hostilities zero homicides

stop extortions robberies and recruitment of children and youths into gangs peace at the

schools) and the government offered to guide public investments in social policies prioritizing

participating municipalities

The truce process was the product of a convergence of vested interests from different sectors

Several earlier attempts at negotiations had failed likely because of a lack of perceived

authenticity on the part of the actors All parties involved in this truce process came to the

table with well-defined group interests and concerns All sides were experiencing fatigue from

the long history of violence Years of fighting deaths and violence had led the mostly

incarcerated first generation those who were still living and who continued to serve as gang

leaders to seek new ways of exercising social power and influence in prison and the

community

The negotiation process In El Salvador the negotiation process between gangs formally began

on March 8 2012 by a decision of the Government to favor the transfer of gang leaders

imprisoned out of the countryrsquos only maximum security prison located in the municipality of

Zacatecoluca That decision was made by the Minister of Justice and Public Security as part of

an agreement between a team of mediators who served as advisors to the Ministry and gang

leaders primarily from MS13 and 18th Street The negotiations were referred to as a ldquotruce

between gangsrdquo and were characterized by the mediators as ldquoa peace processrdquo or a ldquocease of

hostilities between gangsrdquo

Publicly the gang truce was known as an agreement between gangs and not between gangs

and the government because of fear of how those outside the negotiations might interpret the

governmentrsquos actions From the beginning therefore the government never fully

16

acknowledged its participation in the peace process (through representatives) with the gangs

and that reticence was a major impediment throughout the process This was largely because

of the difficulty in communication between the multiple parties which was done through

mediators and necessarily resulted in additional logistical complexity throughout the

negotiations Regardless it is important to recognize that the government played a decisive

and central role in the origin facilitation and promotion of the peace process

In terms of processes interviews with gang leaders indicated that they would communicate

their ldquodemandsrdquo to the mediators who in turn would communicate the gang leaderrsquos demands

to the government The same process was used for the government to communicate with the

gang leaders After an agreement was reached the parties would make proclamations that

emphasized the agreed upon terms of the negotiations These often took the form of public

statements made through local newspapers For example there were more than twenty public

statements made by the gangs where they clearly articulated that their intention was to

reduce the harm that their acts of violence were causing themselves and the communities (For

a detailed discussion of the public statements see Appendix B) In these public statements the

gangs acknowledged the need for their groups to change Both the mediators and the gangs

made clear that the gangs expected the government to respect and respond to their demands

given the good-faith actions that the gangs had already taken

The government on the other hand did not make any formal public statements about their

end of the agreements reached through the negotiations From the outside it appeared that the

government did not develop a formal or consistent policy regarding the truce and instead

assumed that they could achieve a reduction in homicides through the negotiation of increased

prison privileges for gang leaders At the close of the Funes administration the peace process

and negotiations initiated within its framework had reached a plateau There they remained

waiting for the new government to step in and take action as President-elect Salvador Saacutenchez

had announced his support for the truce during the electoral process Sanchez however

withdrew that support when he assumed the presidency in June 2014 Indeed under the

recently elected Sanchez administration the government changed course and became less

17

willing to engage the gangs believing that the government should not negotiate with criminal

organizations

Indeed it should be noted that responses resulting in increased gang influence have been a

concern over the past several decades--from governmental neglect of the problem in the

1990s through the period of the mano dura and super mano dura of the early 1990s to

2000 and to the present gang truce Many have asserted this claim as truth (ie some

responses will increase the influence of gangs) especially in public discourse This study

however does not directly address this point which may or may not be valid9

Strategies employed during negotiations between the gangs and the government Over the

course of the negotiations between the gangs and the gangs and the government at least two

strategies were employed At the beginning of the negotiations strategies were implemented

similar to that of the historic Salvadorian peace accords Each of the parties with the support of

mediators reached agreements and achieved a resolution to their conflict This model

although useful overlooked an important difference between the current violence in El

Salvador and the civil war of the 1980s In a conflict of civilians with the State the legitimacy of

the ldquoadversariesrdquo arises by the need of the state to recover the rule of law It is certainly

possible to reach peace agreements that resolve conflict between gangs but in the current

case with regard to the government there was no legal or policy justification for executing the

truce process On the contrarymdashgovernment negotiations with a criminal group are relatively

rare in modern democratic societies The lack of a legal or policy framework to work from

limited the governmentrsquos ability to be transparent in its response to gangs and may have

undermined its legitimacy with the public

The second strategy employed by the parties was based on reciprocation and cooperation

between actors Early on it was determined by both sides that if one of the parties abandoned

the peace process or did not hold up its end of an agreement the other party would no longer

participate in the negotiation process (Axelrod 1986) In retrospect during the course of the

peace process the gangs implemented agreed-upon terms and positioned themselves as valid

9 We would like to thank one anonymous reviewer for bringing this issue to our attention

18

partners with the government able to negotiate for what is needed by the government in

exchange for what the gang needed10 For example in public statements the gangs insisted

that they had made a goodwill gesture when they declared a unilateral truce and stopped their

involvement in violence The government responded by transferring incarcerated gang leaders

from a maximum-security prison to ordinary criminal prisons11 Next the gangs offered to

disarm the government responded with increased flexibility on control measures at those

prisons This process continued until the demands from both sides grew in a direction that

challenged each sidersquos capacity and willingness to follow through

For instance mediators through the government asked the gangs to end the extortion of

businesses and individuals which are the primary means of subsistence for Salvadorian gangs

The gangs asked that the government reciprocate by eliminating the ldquowitness under a plea

agreementrdquo from the criminal procedural law which is one of the main weapon laws used for

sentencing gang members for complex crimes such as extortion Each of these requests were

more than the parties could carry out Exhibit 1 outlines the different ldquooffersrdquo and ldquodemandsrdquo

made by each party The offers and demands are divided into those that were believed to be

relatively simple and were ldquoofferedrdquo by the party on its own volition and those that were

believed to be more complex and were demanded by the other party

Exhibit 1 Simple demands and concessions made by gangs and the government

Gang concessions Government concessions

Reduce the homicides by more than 50 Transfer the leaders to prisons with lesser security

Hand over 500 firearms to the authorities Allow visits of the children of the gang members

Reduce violent actions at schools Allow night intimate visits

Stop killings at the prisons Allow entry of fast foods

Do not murder custodians police soldiers or their family members

Facilitate the entry into the prison of gang members let out of prison

Give opportunities to some gang members to withdraw

Suspend the searches by the armed forces on persons at the prisons

10 Source httpwwwsalanegraelfaronetes201209cronicas9612 11 This was useful to facilitate the coordination and the communication between the gang members in prisons and in the streets

19

Make it more flexible for new social plans by government to be established in the communities controlled by gangs

Improve the conditions and health assistance at the prisons (eg place tvrsquos to improve mental health of inmates)

Do not burn buses do not kill bus drivers or collecting agents in buses

Allow the Red Cross and journalists to enter into the prisons

The second agenda presented in Exhibit 2 shows the more complex demands that were

not offered by one side or the other but rather were demands placed on the other party These

demands were such that they required a higher level of authority to negotiate in order to

implement the demanded action These demands were considered critical for the peace

process to continue and the delayed responses on both sides stalled the progress of the truce

and led to the parties questioning the legitimacy of the other side all these demands have yet

to be attained The mediators recognized at the end of 2012 that some of the slow progress

was related to the difficulty of making the transition from offers that could be executed

relatively simply to the more complex demands made by each party One of the gang leaders

summarized the issue ldquoWe are not looking for television sets while all of our people continue

living like shit we are not going to try to do everything that is in our power to decrease as much

violence as possible for one television set There are things we are very clear about this [points

to a television set in the room] is a right that the law grantsrdquo

Exhibit 2 Complex demands and concessions made by gangs and the government

Demands by Gangs Demands by Government

Eliminate the figure of witness under a plea agreement from the Criminal Law

Stop homicides and extortions indefinitely

Create international commission to investigate the human rights violation cases of the gang members and their family members on the part of the State (PNC and FA)

Share the information regarding the whereabouts of the brunt of extortion money whether in country or abroad

Institutionalize external and professional surveillance regarding the behavior of the PNC as regards investigations and gang member arrests

Progressive dismantling nation-wide of the clique structures and turfs

Army should stay in its barracks and definitely suspend their participation in public security tasks

Permanently suspend the orders to murder State security and justice agents and their family members (police soldiers judges prosecutors)

20

Clearing of judges prosecutors and police involved in corruption cases against youths in conflict with the law

Permanently suspend the murders extortions and harm caused to public transportation resources and their workers

Maintained a sustained improvement on the conditions of the prisons

Permanently suspend the recruitment of children and youths and hostilities to educational centers

Guarantee working opportunities for the gang members and their family members through specific programs at the municipalities

Share the information about providers of drugs and illegal arms

Television sets were one of the concessions made by the government from the simple list The

gang leader recognized that their intentions in the negotiations extended beyond improving

prison conditions however the government did not have support nor established mechanisms

for carrying out tasks derived from the more demanding list Further as the government

transitioned to new presidential leadership many of the more simple concessions had already

been made and the new government would have to address more complex demands if

negotiations were to continue

The Salvadorian gang truce is remarkable for several reasons First a number of policymakers

and researchers have claimed that the truce saved a large number of lives and was perhaps the

most successful gang truce in the Western Hemisphere Second the Salvadorian gang truce is

somewhat unique in that it might have lasted substantially longer than any other successfully

negotiated gang truce Understanding the temporal impact of the truce is important to

understanding its future potential Third the Salvadorian gang truce is important because a

number of other counties have sought to replicate it For example following the perceived

(and perhaps real) success of the Salvadorian gang truce the nations of Honduras Belize and

Guatemala instituted or tried to institute a similar type of truce Understanding the impact of

the Salvadorian gang truce will further help us understand the potential impact of such a

process on violence in these other nations

Methods

We examined the impact of the gang truce by merging four separate data sets First data from

the 2007 El Salvador Population and Housing Census provided municipal level measures of

21

percent unemployed percent male aged 10-29 percent female headed households ethnic

composition in-migration income percent urban percent households rented and percent

who had a high school education These data were obtained directly from El Salvadors Ministry

of Economy Second municipal level crime data (ie homicide extortion rape theft robbery

and auto theftrobbery) were provided by the National Civil Police (PNC) by month and year for

the period between January 1 2010 and June 30 2014 Disappearance data was also provided

by the police aggregated by year and municipality Third police gang intelligence data was

provided by the National Civil Police (PNC) Specifically we received the number of police

recorded MS 13 and 18th Street gang members by municipality in 2011 the latest year for

which these data were made available Last we acquired 2011 prison gang intelligence data on

the number of imprisoned MS 13 and 18th Street members by municipality from the

Salvadorian National Bureau of Prisons All four data sources were linked together using a

unique municipality identification number and month Collectively they provide an

opportunity to examine the impact of the gang truce in El Salvador and understand several

competing explanations for any changes that might have occurred

The geographic unit of analysis for the present study is the municipality According to the 2007

El Salvador Population and Housing Census there were 14 departments divided into 262

municipalities (the equivalent of a county in the USA) in El Salvador Of the 262 municipalities

ten (38) of the municipalities were eliminated from the analysis because of missing data

Measures

Dependent variables The dependent variable in the present study is the monthly number of

homicides in each El Salvadoran municipality expressed as the number of homicides in

municipality i at time t Homicide data were collected from January 1 2010 through June

30 2014 for a total of 54 months This includes data for a period of 26 months prior to the gang

truce and 28 months following the implementation of the gang truce As shown in Exhibit 3

there were a total of 14148 homicides over the study period with each municipality averaging

371 homicides per month

yit

22

Exhibit 3 Descriptive statistics

n mean sd

Urban Opportunity Factor 252 111 95

male 10-29 years old 252 1960 171

female-headed household 252 3433 499

unemployed 252 1177 686

Racialethnic heterogeneity 252 17 13

of prisoners MS13 252 1579 3960

of prisoners 18th Street 252 1268 4096

of prisoners MS13_spatial weight 252 1579 3960

of prisoners 18th Street _spatial weight 252 1268 4096

MS13 on the street 252 4146 10646

18th Street on the street 252 2513 8622

Violence free municipality 252 04 20

Truce 14148 371 783

Monthly homicide rate per 1000000 14148 51 49

Independent variables Several independent variables were included in our model Our

measure for the intervention is a dummy variable in which the value of 0 is used to represent

the pre-intervention months and 1 represents the post-intervention months We also

included a dummy variable for each community that participated in the violence-free

municipality initiative (sanctuary municipalities) to address issues of external validity (eg

multiple treatment inference) In other words municipalities that participated in the

violence-free initiative might have experienced or participated in something that either

enhanced or detracted from the impact of the larger gang truce The initiative called by some

the second phase of the gang truce took place about 8 to 9 months after the original gang

truce was negotiated in March 2012 In participating sanctuary municipalities gang members

agreed to stop violence and crime in exchange for a reduction in police operations and night

raids12 The municipalities that participated in the initiative included Santa Tecla

Quezaltepeque La Libertad Ilopango Ciudad Delgado Apopa Sonsonate San Vicente

Zacateatecoluca Puerto El Triunfo and Nueva Concepcioacuten

12 See httpwwwcispesorgblogviolence-free-cities-inaugurated-as-second-phase-of-gang-truce

Tt

Si

23

We included several independent variables related to the presence of gangs by gang for each

municipality Specifically we included count variables on the number of MS13 and 18th Street

gang members who were on the streets and in prison by municipality As noted before the

gang truce was made between a relatively small number of imprisoned gang leaders from both

MS13 and both fractions of 18th Street who agreed to stop street-level gang violence on the

condition that they would be transferred to lower security prisons and granted special

privileges

We wanted to understand the impact of the truce in the context of gang presence

Municipalities with high numbers of MS13 and 18th Street members whether they were in

prison or on the streets should have experienced a greater reduction in homicides because of a

greater span of control over these communities Accordingly municipalities with low numbers

of MS13 and 18th Street members should have experienced a lesser reduction in homicides

because of a limited span of control over these communities As Exhibit 3 (above) shows on

average municipalities had about sixteen MS13 and thirteen 18th Street members in prison

and forty-one MS13 and twenty-five 18th Street members on the street We examined whether

gang members in El Salvador were randomly distributed and found that there was spatial

clustering in the number of MS13 and 18th Street members who were in prison In order to

control for the clustering we included the spatial lag (weight) of the MS13 and 18th Street

members in prison

Control variables A series of control variables known through prior research to be related to

violence in communities were added using the census data We began with municipal-level

measures of percent unemployed percent male aged 10-29 percent female-headed

households ethnic composition in-migration income percent urban percent households

rented and percent with a high school education Initial analysis found that five community

variables were highly correlated and loaded on the same factor As seen in Exhibit 4 these

included the percentage of persons who had moved there from another municipality (in-

migration) average income in the municipality percent of the population living in an urban

area percent of houses that are rented and percent of residents with at least a high school

education We labeled this factor Urban Opportunity

24

Exhibit 4 Factor loadings for urban opportunity

Variables Factor

loadings

In Migration Percent of population moved in from another municipality 0638

Income Average monthly income per household (colones) 0886

Urban Percent of population that is urban 0845

Rented Percent of households rented 0761

Education Percent of residents who have at least a high school education 0742

Percent of variance 6071

Eigenvalue 341

Extraction method Principal Axis Factoring

We also calculated a measure of ethnic-heterogeneity from relevant census data Ethnic

heterogeneity which varies from 0 to 1 was calculated by taking one minus the squared

proportions of the population in each ethnic group (White Mestizo other) As with the

percentages of male population that is 10-29 years old female-headed households and

unemployed persons ethnic heterogeneity has consistently been associated with violence in

general (Kubrin and Weitzer 2003) and with gang violence in particular (Katz Maguire and

Roncek 2002 Rosenfeld Bray and Egley 1999)

Analytic Strategy

National-level analysis Our analysis began with a simple t-test of the mean number of

homicides at the national level before and after the truce providing the most basic omnibus

test of an effect We then presented our time series model to estimate the effect net of

seasonality and temporal trends This analysis included a set of simple ARMA (autoregressive

and moving average) models with two lag periods and one period of a moving average Initial

analysis of the number of homicides by month indicated that partial temporal autocorrelations

existed for two lags The first model used only data prior to the truce This model included a

linear time trend variable and dichotomous variables for each month (except January) We used

this model to forecast the expected number of homicides for the truce period This series of

analyses was for illustrative purposes only as the number of data points used in the forecast

25

was too small to provide meaningful confidence intervals The second ARMA model employed

all data from January 2010 through June 2014 and included a dichotomous indicator for the

truce period The effect of this variable was the average change in the number of homicides

net of seasonality (months) and temporal trends

Municipality multilevel models The analysis at the municipal level presented four challenges to

a typical regression model First since our outcome has a highly skewed distribution with

varying levels of exposure (population) we employed a generalized linear model to capture the

correct distribution In this case we employed a negative binomial distribution rather than a

Poisson model due to the over-dispersion in our outcome created by analyzing monthly

municipal datamdashthat is there were several months and municipalities where no or very few

homicides occurred

The second challenge was our need to measure the between-municipality variation of pre-truce

homicide rates and the program effect A fixed-effects model would have been inappropriate

because it would not have allowed us to estimate these variance components Thus we

employed a multilevel random effects generalized linear model

The third challenge was temporal autocorrelation Analysis at the national level indicated

partial autocorrelations in the first and second lags Although a generalized model does not

allow for auto-correlated residuals we addressed this by entering in the model two lags of the

homicide rate

The fourth and final challenge to this estimation was spatial autocorrelation An examination of

the Moranrsquos I and autocorrelation coefficients (exhibit not shown) indicated a low level of

autocorrelation for each month but many months were still statistically significant Thus we

estimated spatial lags of the homicide rate and entered them into the model

Accordingly we analyzed the data using random effects negative binomial models regressing

the rate of homicide on various months on predictors with both temporal and spatial lags For

clarity we present the model using HLM (Bryk et al 1996) notation At the first level (time) we

estimated the log number of homicides using a negative binomial distribution with predictors

26

that included the truce period calendar month linear month trend a

one period lag of the homicide rate a two-period lag of the homicide rate a

spatial lag of the homicide rate and an over-dispersion parameter which has a

Gamma distribution of Thus the final level 1 model is

At the municipality level (level 2) the intercept is a function of the presence of the

Violence Free Municipality program the log number of MS prisoners the log

number of 18th Street prisoners the log of the spatial lag number of MS prisoners

the log of the spatial lag of the number of 18th Street prisoners the log of the

number of MS members the log of the number of 18th Street prisoners the

control variables detailed above and the random effect Thus the level 2 model for the

intercept is

Also at the municipality level (level 2) the effect of truce is a function of SAFE the

log number of MS prisoners the log number of 18th Street prisoners the log of

the spatial lag number of MS prisoners the log of the spatial lag of the number of 18th

Street prisoners the log of the number of MS members the log of the number

of 18th Street prisoners and the random effect Thus the level 2 model for the

truce effect is

TRUCEt mt TIMEt

HRt-1i HRt-2i

HRti vti

Gamma1

aa

aelig

egraveccedilouml

oslashdivide

mti = expb0 i + b1iTRUCEt + lmmt

m=1

11

aring + l12TIMEt

+l13HRt-1i + l14HRt-2i + l15HRti + vti

eacute

euml

ecircecircecirc

ugrave

ucirc

uacuteuacuteuacute

exp vti[ ] ~ Gamma1

aa

aelig

egraveccedilouml

oslashdivide

b0 i

SAFEi PMSi

P18thi

PMSi P18thi

MMSi M18thi

Ci u0 i

b0 i = g 00 +g 01SAFEi +g 02 ln PMSi +1( ) +g 03 ln P18thi +1( ) +g 04 ln PMSi +1( )

+g 05 ln P18thi +1( ) +g 06 ln MMSi +1( ) +g 07 ln M18thi +1( ) + pC

C

aring Ci + u0i

b1i SAFEi

PMSi P18thi

PMSi

P18thi MMSi

M18thi u1i

27

The key parameters of our analysis are in this expression The truce effect is dependent on

the log number of MS prisoners among other characteristics Thus the percent

reduction in homicides can be calculated from this model using the following formula

which we plot below (see Quantitative Findings below)

To answer questions about the variation in the effects of the gang negotiations we predicted

best linear unbiased predictions (BLUP) of the Truce random effect which estimates the

between-community variance of the immediate effect of the truce Since we assumed that the

interventions were not evenly spread throughout communities in this study the mixed model is

appropriate to address this issue We conducted this analysis for homicide using a Stata

generalized linear mixed model using full maximum likelihood and an identity covariance matrix

of random effects

Findings

Exhibits 5 and 6 present our findings related to the number of police recorded homicides in El

Salvador from January 2010 through June 2014 The trends in the data and the results of the t-

test show that in the 26 months prior to the gang truce there were on average about 354

homicides per month compared with about 218 homicides per month in the 28 months

following the gang truce Exhibit 7 shows that the gang truce resulted in 5501 fewer homicides

than otherwise would have occurred

b1i = g 10 +g 11SAFEi +g 12 ln PMSi +1( ) +g 13 ln P18thi +1( ) +g 14 ln PMSi +1( )

+g 15 ln P18thi +1( ) +g 16 ln MMSi +1( ) +g 17 ln M18thi +1( ) + u1i

b1i

ln PMSi +1( )

1- exp g 10 + ln PMSi +1( )g 12( )eacute

eumlugraveucirc( )acute100

u1i

28

Exhibit 6 Number of homicides in El Salvador by month with ARIMA Model Fit

Exhibit 5 Change in homicides from pre-truce to post-truce

Obs Mean SD 95 confidence interval

Pre-truce period 26 35442 4200 33745 37138

Post-truce period 28 21835 6361 19369 24302

Change -13606 10638 16574

=p lt 05

29

Exhibit 7 Forecast of homicides without gang truce

Next as seen in Exhibit 8 we examined the spatial distribution of the change in the homicide

rate after the implementation of the gang truce The analysis showed that of the 252 analyzed

municipalities 243 (93) experienced a decrease in homicides however within these

municipalities there were wide variations in the degree of the decline For example of the 243

municipalities that experienced a decrease the decrease in the homicide rate varied from

about 59 percent of these municipalities experiencing a 1-74 percent decrease to about 9

percent of municipalities experiencing a 75 percent or higher decrease Additionally a modest

number (n=19 7) of municipalities experienced an increase in their homicide rate

30

Exhibit 8 Percent reduction in homicide rate by number of municipalities

Exhibit 9 presents six negative binomial models for the monthly number of homicides The first

model examines the impact of the gang truce and implementation of the Violence Free

Municipality program and our interaction variable that measures the additive effect of both

the gang truce and the Violence Free Municipalities program on the number of homicides We

found that although the gang truce was associated with a significant decline in homicides the

Violence Free Municipality program was related to a significant increase in homicides13

However countrywide the additive effect of implementing the gang truce and the Violence

Free Municipalities program was associated with a significant decline in homicides

13 To be clear our bivariate analyses showed that the violence free municipalities program was related to a significant decline in homicides For example on the one hand those municipalities that did not participate in the violence free municipalities program experienced a decline in their homicide rate from 406 homicides per 100000 population in the pre-truce period to 321 homicides per 100000 in the post-truce period On the other hand those municipalities that participated in the violence free municipalities program experienced a substantially greater decline in their homicide rate from 752 homicides per 100000 population in the pre-truce period to 392 homicides per 100000 in the post-truce period However after other variables are controlled for in our negative binomial models we found that the violence free municipalities program did not have a positive impact on homicides over and above the gang truce itself

31

Exhibit 9 Negative Binomial models for monthly number of homicides

Model 1 Coefficient (se)

Model 2 Coefficient (se)

Model 3 Coefficient (se)

Model 4 Coefficient (se)

Model 5 Coefficient (se)

Model 6 Coefficient (se)

Homicide - 1 month lag 0010 0010 0010 0010 0010 0010

(0002) (0002) (0002) (0002) (0002) (0002)

Homicide - 2 month lag 0012 0013 0013 0013 0013 0013

(0002) (0002) (0002) (0002) (0002) (0002)

Homicide rate spatial lag 0003 0003 0003 0003 0003 0003

(0001) (0001) (0001) (0001) (0001) (0001)

Gang truce implemented -0483 -0541 -0553 0057 -0544 0564

(0066) (0065) (0064) (0278) (0064) (0279)

Violence free municipality 0497 0183 -0003 -0076 -0037 -0094

(0181) (0167) (0152) (0143) (0153) (0145)

Ln( +1) of MS13 prisoners 0066 0176 0062 0171

(0023) (0028) (0043) (0028)

Ln( +1) of 18th St prisoners 0067 0044 0064 -0043

(0021) (0024) (0021) (0024)

MS13 prisoner spatial lag -0004 0054 -0003 0049

(0045) (0055) (0046) (0056)

18th St prisoner spatial lag 0022 0014 0019 0012

(0027) (0032) (0027) (0032)

Gang truce violence free municipality

-0346 -0294 -0309 -0161 -0304 -0059

(0158) (0156) (0158) (0146) (0158) (0145)

Gang truce Ln( +1) of MS13 prisoners

-0185 -0185

32

(0030) (0030)

Gang truce Ln( +1) of 18th St prisoners

0160 0060

(0025) (0025)

Gang truce Ln( +1) of MS13 prisoners spatial lag

-0139 -0138

(0064) (0064)

Gang truce Ln( +1) of 18th St prisoners spatial lag

0023 0023

(0039) (0039)

Ln( +1) of MS13 on street 0023 0018 0021 0016

(0013) (0012) (0013) (0012)

Ln( +1) of 18th St on street 0043 0039 0040 0037

(0016) (0015) (0016) (0015)

Time 0010 0011 0011 0011 0011 0011

(0002) (0002) (0002) (0002) (0002) (0002)

1month 0064 0065 0067 0067 0067 0067

(0055) (0055) (0055) (0055) (0055) (0055)

2month -0020 -0020 -0022 -0023 -0022 -0023

(0056) (0056) (0056) (0056) (0056) (0056)

3month 0088 0088 0188 0088 0088 0088

(0052) (0052) (0052) (0052) (0052) (0052)

4month -0011 -0011 -0011 -0011 -0011 -0011

(0052) (0052) (0052) (0052) (0052) (0052)

5month 0020 0020 0020 0021 0020 0021

(0052) (0052) (0052) (0052) (0052) (0052)

6month 0000 0000 0000 0000 0000 0000

7month 0036 0036 0036 0036 0036 0036

(0052) (0052) (0052) (0052) (0052) (0052)

33

8month 0047 0047 0047 0046 0047 0046

(0051) (0051) (0051) (0051) (0051) (0051)

9month -0055 -0055 -0057 -0058 -0057 -0058

(0053) (0053) (0053) (0053) (0053) (0053)

10month 0040 0040 0041 0041 0041 0041

(0052) (0052) (0052) (0052) (0052) (0052)

11month 0022 0022 0022 0022 0022 0022

(0052) (0052) (0052) (0052) (0052) (0052)

12month -0040 -0040 -0042 -0043 -0042 -0043

(0053) (0053) (0053) (0053) (0053) (0053)

Urban opportunity factor 0209 0042 0033

(0040) (0040) (0039)

male 10-29 0039 0007 0009

(0021) (0019) (0019)

female-headed household -0006 0002 0000

(0007) (0007) (0006)

unemployed 0001 -0002 -0000

(0006) (0005) (0005)

Racialethnic heterogeniety 0058 -0039 -0087

(0263) (0235) (0230)

Intercept -10571

-10544

-11054 -11618 -11015 -11558

(0064) (0060) (0194) (0241) (0212) (0257)

Ln(Alpha) -2287 -2289 -2293 -2294 -2293 -2294

(0113) (0113) (0114) (0114) (0114) (0114)

Truce random effect 0202 0193 0202 0144 0201 0143

(0039) (0038) (0040) (0031) (0040) (0031)

Intercept random effect 0315 0226 0192 0164 0183 0158

(0047) (0035) (0032) (0027) (0031) (0027)

Cov(Truce and Intercept) -0162 -0099 -0122 -0086 -0114 -0090

(0037) (0031) (0031) (0025) (0031) (0026)

34

The second model includes those variables from Model 1 but also includes variables that

controlled for community-level structural factors In this model the gang truce remains

significantly associated with a decline in homicides but the Violence Free Municipality program

and our interaction variable were no longer related to a reduction in homicide Although the

percentage of residents in a municipality who are male aged 10 to 29 female-headed

households percent unemployed and ethnic heterogeneity were unrelated to changes in

homicide the urban opportunity factor was significantly associated with homicide

Models 3 and 4 included our measures of intervention as well as our measures of the number

of MS13 and 18th Street on-the-street and incarcerated gang members at the municipality level

The analysis showed that while the number of MS13 on the street was unrelated to homicides

the number of 18th Street gang members was associated with an increase in homicide These

analyses also indicated that the number of incarcerated MS13 and 18th Street gang members

from a municipality was associated with a significant increase in homicides We further

examined whether this relationship was associated with the gang truce by including two

interaction variables as presented in Model 4 One measured the interaction between the

number of incarcerated MS13 gang members and the gang truce and a second measured the

interaction between the number of incarcerated 18th Street gang members and the gang truce

The results showed that following the gang truce the number of incarcerated MS13 gang

members was related to a significant decrease in homicides and the number of incarcerated

18th Street gang members was related to a significant increase in homicides

Models 5 and 6 in Exhibit 9 present the same two models (3 and 4) as above but include the

community-level structural variables Model 5 once again shows that the number of MS13

members on the street remains unrelated to the change in homicides and the 18th Street

members on the street was associated with a significant increase in homicide However Model

6 shows that following the gang truce the number of incarcerated MS13 gang members

remained significantly associated with a decline in homicides while the number of incarcerated

18th Street gang members remained significantly associated with an increase in homicides

35

Community-level structural factors were found once again to be unrelated to changes in

homicide rates

Exhibit 10 below further illustrates the relationship between reductions in homicides by

municipality and the municipal-level presence of MS13 and 18th Street gang members in

prison The figure shows that in municipalities with no incarcerated MS13 and 18th Street gang

members no change in homicides occurred following the gang truce However when a

municipality had ten MS13 gang members imprisoned on average that municipality

experienced a 55 percent reduction in homicides When a municipality had fifty MS13 gang

members imprisoned those municipalities on average experienced a 71 percent reduction in

homicide following the truce Conversely our analysis showed that the number of 18th Street

gang members in prison (from a municipality) had a significant and positive impact (that is the

number increased) on homicides following the truce For example if a municipality had ten

18th Street gang members imprisoned on average that municipality experienced a 31 percent

increase in homicides14

14 We examined whether there was an interaction effect between gang members on the street following the gangs truce Analysis for both MS13 and 18th street and their interaction with the gang truce showed no effect

36

Exhibit 10 Impact of the gang truce by number of gang members in prison at the municipal level

The Gang Truce and Crime Displacement

We examined two types of displacement crime type displacement and method displacement

Some policymakers have suggested that although homicides may have declined as a

consequence of the gang truce other forms of criminality or categories of reported crime may

have subsequently increased Crime type displacement occurs when offenders who focus on

one type of crime switch to another form of crime in order to avoid detection or to benefit in

some other way (Eck 1993) In El Salvadorrsquos case some critics of the gang truce have suggested

that as homicides decreased other forms of criminality such as extortion might have

increased substantially due to gang membersrsquo increased freedom to conduct activities inside

and outside of prison (Dudley 2013 Parkinson 2014) Method displacement occurs when

offenders change their tactics or methods of conducting crimes as a consequence of an

intervention (Eck 1993) Following the gang truce in El Salvador some analysts argued that gang

37

members might have begun to hide the bodies of homicide victims to avoid detection and to

ldquoprotect the integrity of the countryrsquos gang trucerdquo (Bargent 2013 1)

As noted above we rely on police data that measured monthly numbers of thefts extortions

robberies rapes and auto theftsrobberies by municipality We also received data on the

annual number of disappearances in each municipality by year Exhibit 11 presents the

descriptive statistics for these variables It shows that the rates of theft and robbery did not

change significantly between the pre- and post-true periods and the rate of extortions

significantly declined The t-tests showed that the rates of rape disappearance and

theftrobbery fromof an auto increased significantly

Exhibit 11 Descriptive statistics for measures of displacement

Mean Std Err [95 Conf Interval]

Theft rate

Pre-truce 1238 049 1142 1335

Post-truce 1213 049 1115 1310

Extortions rate

Pre-truce 353 023 308 398

Post-truce 271 021 229 314

Robbery rate

Pre-truce 544 025 495 593

Post-truce 553 025 504 602

Rape rate

Pre-truce 049 003 043 056

Post-truce 064 005 054 073

TheftRobbery Vehicle rate

Pre-truce 132 022 090 175

Post-truce 165 023 120 210

Disappearance rate

Pre-truce 805 068 670 940

Post-truce 1007 064 882 1132

p lt 05

We further examined the data (excepting disappearance data) similarly to the analyses above

in which we used random effects negative binomial models and regressed the number of

38

crimes (ie theft extortion robbery rape and auto theftrobbery) on various months on

predictors with both temporal and spatial lags Our independent and control variables

remained the same as those used in Model 6 (displayed above in Exhibit 9) We found that

over the study period there was no significant change between the pre-truce and post-truce

periods in the number of thefts extortions robberies rapes and auto theftsrobberies (tables

not shown)

As noted above we also received data on the number of disappearances by municipality and

year but because the data were provided by year there were not enough data points to

examine them temporally Therefore we added this covariate (number of disappearances by

municipality and year) to Model 6 in Exhibit 9 The results of the analysis indicated that the

number of disappearances was not significantly related to change in homicides the gang truce

remained associated with a significant reduction in homicides and our interaction variables

(number of incarcerated MS13 amp 18th Street members truce) remained significantly

associated with homicides

Conclusions

The present study sought to examine the impact of the gang truce on violence and other forms

of criminality We examined homicide data by municipality prior to and after the gang truce

Our outcome variables were obtained from the PNC along with several independent and

control variables obtained from the Ministry of Economy and the National Bureau of Prisons

We analyzed this data using a number of time series and random effects negative binomial

models where we regressed the rate of homicide on various months on predictors with both

temporal and spatial lags and controlled for other contextual factors This type of analysis

allowed us to correct for the number of homicides in one month being related to the number of

homicides in the previous month (ie temporal autocorrelation) Additionally it allowed for

the fact that some communities participated in supplemental interventions (ie violence free

municipalities initiative) which might have impacted the outcome in the same way in a given

month These techniques allowed us to isolate the effects of the gang truce as best possible

We also used data from the PNC that measured by municipality the number of thefts

39

extortions robberies rapes auto theftsrobberies and disappearances to examine the

possibility of the trucersquos impact on crime displacement and diffusion of benefits`

Our findings suggest that El Salvadorrsquos gang truce had a significant and dramatic impact on the

homicide rate The mean number of monthly homicides declined from about 354 prior to the

truce to about 218 following the truce for a net decrease of about 136 homicides per month

Our forecast showed that between March 2012 and June 2014 the truce had saved about 5501

lives As will be discussed further below the analysis suggests that key stakeholders have the

capacity to renegotiate existing norms of violence and that at least some gangs have the

capacity to exert substantial informal social control over their members that can result in

reduced violence

The gang truce also lasted substantially longer than previously evaluated truces Ordog et al

(1995) for example reported that the much publicized gang truce in Los Angeles decreased

homicides by about 35 percent for the first three months but then doubled in months four

through eleven compared with the pre-truce period Similar findings were reported in Trinidad

and Tobago (Maguire Katz and Wilson 2013) where it was determined that homicides

declined for a brief period of time (again for about three months) but then increased

substantially over the long term (12 months) These results suggested that gang truces may

produce short-term benefits yet result in long-term adverse consequences

The findings of the present case study suggests that some gang truces might last longer than

previously believed While the number of homicides began to slowly increase about 12 months

following the truce in El Salvador the results clearly showed that almost two years following

the truce homicides still remained below those experienced prior to the truce

We found however that the gang truce did not result in a homogenous decline in violence

across municipalities About 61 percent of municipalities experienced a decline in homicides

but the decline in violence varied substantially between municipalities For example while

about 16 percent of these municipalities experienced a 25 percent reduction in homicides a

number of others (37) experienced a 75 percent or greater reduction Furthermore it is

important to note that about 20 percent of municipalities experienced a modest increase in

40

homicides This suggests that the impact of a gang truce might be variable and could be

dependent on contextual factors We examined the possible influence of these factors by

assessing the impact of social structural factors and the presence of gangs on the municipal-

level impact of the gang truce While we found that social structural factors were unrelated to a

decline in homicides municipal-level gang presence was associated with the decline in violence

as a consequence of the gang truce

We examined this issue further by parsing out the relative influence of the number of MS13

and 18th Street gang members on the street and in prison from each municipality As noted

above we hypothesized that municipalities where gang member presence was high regardless

of their presence on the street or in prison would experience a greater reduction in homicides

because of their increased influence in these areas We found however that the relationship

was not as direct as we would have suspected In particular our analyses indicated that

following the truce the number of MS13 and 18th Street gang members on the street in a

municipality was not significantly related to a decline in homicide but the number of

imprisoned MS13 and 18th Street gang members was associated with a significant change in

homicides following the gang truce

Of special note was our finding of the differential impact of the truce based on gang Although

we found that the number of imprisoned MS13 gang members was associated with a significant

decline in homicides following the gang truce we also found that the number of imprisoned

18th Street members was associated with a significant increase in homicides following the truce

These divergent findings might be associated with each gangrsquos organizational structure and its

capacity to regulate member behavior

Much prior research suggests that imprisoned MS13 gang members have substantial influence

over violence in Salvadoran communities15perhaps even more so than formal mechanisms of

15 Instituto Universitario de Opinioacuten Puacuteblica (Iudop) La situacioacuten de la seguridad y la justicia 2009-2014 entre expectativas de cambio mano dura militar y treguas pandilleras Aguilar Jeannette (Coord) [et al] --1ordf ed --San Salvador El Salv Instituto Universitario de Opinioacuten Puacuteblica (Iudop) 2014- 260 p graacuteficos tablas cuadros y figuras 28 cm (Talleres graacuteficos UCA) httpwwwucaedusviudopwp-contentuploadslibro_la_situaciC3B3n_de_la_seguridadpdf

41

social control such as the police and courts16 Law enforcement officials for years have claimed

that MS13 is one of the most organizationally sophisticated street gangs in the Western

Hemisphere17 The gang has been widely characterized as having a highly vertical organizational

structure and strong control over criminal enterprises in gang-controlled neighborhoods and as

being decidedly capable of enforcing rules through discipline MS13 leadership resides in the

nationrsquos prison system The Ranfla (gang leadership) is comprised of thirteen MS13 gang

members (PNC 2011) who direct coordinate and authorize street crime and other gang activity

from prison Our findings lend support to the idea that MS13 is more organized than the typical

street gang and that imprisoned MS13 gang members exhibit strong influence over their fellow

gang members on the street

Our findings also suggest however that the gang truce had a boomerang effect in

municipalities with high numbers of imprisoned 18th Street members implying that 18th Street

might not have as much organizational capacity to regulate violence on the streets as MS13

The truce provided incarcerated MS13 and 18th Street gang leaders an opportunity to

negotiate with high-ranking officials and influential diplomats This may have increased their

legitimacy inside and outside of their gangs It appears that MS13 was able to exert its span of

control over the communities in which they had influence and they were able to deliver on the

terms of the gang truce negotiations In the case of 18th Street however incarcerated gang

members may not have had the same organizational capacity for communicating and carrying

out directives In fact a review of the gang truce indicated that there was a conflict taking place

between two fractions within 18th Street Consequently the organizational structure and

culture of 18th Street might be more diffuse than that of MS13 and its leadership structure

might not be as strong because of the internal fractures within the gang This might further

explain why homicides increased in 18th Street communities The internal fractures within the

18th Street gang may have resulted in intra-gang violent conflict that was largely contained

16 Instituto Universitario de Opinioacuten Puacuteblica (Iudop) La situacioacuten de la seguridad y la justicia 2009-2014 entre expectativas de cambio mano dura militar y treguas pandilleras Aguilar Jeannette (Coord) [et al] --1ordf ed --San Salvador El Salv Instituto Universitario de Opinioacuten Puacuteblica (Iudop) 2014- 260 p graacuteficos tablas cuadros y figuras 28 cm (Talleres graacuteficos UCA) httpwwwucaedusviudopwp-contentuploadslibro_la_situaciC3B3n_de_la_seguridadpdf 17 httpwwwlaprensagraficacomeua-declara-a-ms-grupo-delictivo-transnacional

42

within 18th Street controlled territories Further analysis is needed to examine this specific

issue

While not the primary focus of this case study we controlled for any impact that the Violence

Free Municipalities program might have had on homicide in 11 municipalities during the truce

As noted above the Violence Free Municipalities program served as the second phase of the

gang truce (Cawley 2013) The program was first proposed by the two mediators and designed

by the Technical Committee on Violence and Crime Reduction which included representatives

from OAS MJPS the mediators and the Humanitarian Foundation (CISPES 2013) The program

involved the mayors of each municipality collaborating with gang leaders to design prevention

and intervention resources for gang members and at-risk youth The Minister of Justice and

Security David Mungia Payes announced that his ministry would facilitate $74 million in

funding from the OAS UN and other donors to implement the programing In exchange gang

leaders agreed to end violence and other criminality in the Violence Free Municipalities (CISPES

2013) The negotiators also agreed to further discuss both gangsrsquo demand to repeal of the

2010 law that increased the capacity of the police and prosecutors to crack down on gangs

(Ayala 2014)

Our findings suggested however that the Violence Free Municipalities program was unrelated

to change in homicides in these communities While much additional research is needed to

understand why the program was not effective it might have been because the gangs had

already agreed to a truce and had already reduced violence to the extent that they could

Conversely the demands made by the gangs as part of the Violence Free Municipalities

program might have been more than could be delivered The time and resources required to

implement the programming and the political capital that was required to repeal legislation

might have been much more than could be delivered Future research is needed to examine the

processes and impact of the Violence Free Municipalities program

Regardless the base rate of violence in El Salvador was reset for a period of time suggesting

that perhaps the gang truce substantially altered existing norms of violence Klein (1995)

argued that cycles of gang violence (that is perceived or real changes in gang activity) are

43

typically the consequence of seasonality epochal variation (peaks and valleys in violence) and

illusion (it appears as if there is a real change in violence but there is not) In El Salvador we

appear to have observed a self-directed cycle of normative change wherein incarcerated MS13

gang leaders directed a reduction in violence by actuating their organizational span of control

through the gang truce Some of the most powerful and influential gang members in El Salvador

used their political social and economic capital to promote the truce and articulate new norms

of violence They were able to do this by leveraging their informal social control over the streets

through actual or perceived threats of violence against those who violated the terms of the

truce

However following a change in government leadership and the government subsequently

distancing itself from the gang truce the conditions and capacity of MS13 leaders to intervene

in local violence might have deteriorated and violence began to increase substantially This is in

part because the third parties in the negotiation were no longer able to communicate with

government officials about furtherance of the truce Specifically the mediators were no longer

able to negotiate on the behalf of the government and were no longer permitted entrance to

the prisons where they could negotiate with gang leaders

44

Case Study B Gang Trucemdash The Jamaican Experience

In the latest report by the UN Office of Drugs and Crime Jamaica ranked as the sixth most

violent nation in the world with a homicide rate of about 521 per 100000 residents (UNDOC

2011) Policymakers and researchers have attributed the nationrsquos high level of violence to such

factors as drug trafficking (Klein Day and Harriott 2004) access to illegal firearms (Lemard and

Hemenway 2006) and historical processes that include a legacy of conflict between the

nationrsquos two primary political parties (Sives 2002 Figueroa and Sives 2003 Moser and Shrader

1999) all of which have facilitated the entrenchment of the more powerful gangs in

communities of the urban poor Likewise high levels of income inequality and chronic youth

unemployment (Francis and Lyare 2006) problematic urbanization (Stone 1975) social

marginalization and an emergent subculture of violence (Harriott 2008) have contributed to the

nationrsquos violence problem Whatever the causes violence has had a considerable impact on

Jamaicarsquos social and economic development it has decreased investor confidence (Schwab and

Porter 2008) tourism (Harriott 2007) and access to public services It has also increased the

costs related to the health care system (Mansinghand and Ramphal 1993) the criminal justice

system (Caribbean Human Development Report 2012) and the education system (Moser and

Holland 1997)

Jamaicarsquos homicide problem is closely associated with its gang problem The Jamaican

Constabulary Force (JCF) has estimated that some 272 gangs are active in the nation most in or

near the capital city of Kingston (Harriott 2014) Gang types and their respective historical

patterns of conflict matter in Jamaica as these variations may determine their predisposition or

amenability to lasting rather than opportunistic truces Jamaican gangs include territorially

organized crime groups conflict gangs defense crews who regard themselves as defenders of

their communities (Levy 2009) and other less cohesive more transient territorial groups Some

of the latter are predatory others bond around the taken identities that generate conflicts of

other similar groups18

18 These are generally rooted in subcultural issues such as the demand to be treated with respect or with the indiscretions and self-centered aggressiveness of individual members that may be associated for example with sexual competition for the favors of women

45

Estimates of the gang-related homicide rate in Jamaica vary perhaps because the crime is not

clearly defined as far as attributing a death to a gang Regardless researchers agree that the

proportion of homicides that are gang related has increased substantially For example Harriott

(2003) reported a fourfold increase in the rate of gang-related homicides between 1983 and

1997 Likewise Hill (2013) using official police data found an eightfold increase from 2001 to

2009 reportedly in 2001 only about 64 percent (n=73) of the nationrsquos 887 homicides were

gang related compared with 523 percent (n=879) of its 1682 homicides in 2009

To address the problem Jamaica has initiated traditional law enforcement strategies such as

establishing a specialized gang unit (Sinclair 2004) initiating curfews (Sinclair and Tuner 2005)

declaring states of emergency (Jamaican Observer 2010) and implementing community-

oriented policing (Kolpack 2006) It also attempted legislative reforms to curb election fraud

and electoral-related violence that involved local gangs (Levy 2009) None of these strategies

stemmed the tide of gang violence In 2002 the Minister of National Security established the

Peace Management Initiative (PMI) (Henry 2011) to augment governmental and non-

governmental organizational capacity to settle gang disputes in the community through

intervention-based programming such as ceasefires and gang truces

The current study examines the peace initiative instituted by the PMI in Greater August Town

Our objective was to understand the negotiation processes undertaken with and between

gangs and other stakeholders Among other things we were interested in identifying the actors

involved in the negotiations the goal(s) of the negotiations and the strategies employed to

carry them out Most importantly however we wanted to determine whether the gang truce

resulted in the desired outcome a reduction in the number of homicides in the Greater August

Town area

The Greater August Town (Jamaica) Peace Initiative

Greater August Town is located on the northeastern outskirts of the city of Kingston The low-

income area has high rates of youth unemployment and a history of gang-related violence

(Charles 2004 Levy 2009) Over the last decade the Greater August Town community has

sought improved living conditions and revitalization (Levy 2009 95) The arearsquos inherent

46

resilience has been augmented by nearby intellectual and cultural engines such as the

University of West Indies the University of Technology and University Hospital (Charles 2004

38)

Greater August Town is comprised of the communities of August Town proper (which is

fractured into several locales with gang-given names such as ldquoVietnamrdquo and ldquoOpen Landrdquo and

city government-approved names such as Hermitage Goldsmith Villa (Angola) and Bedward

Gardens These socially defined community divisions and subdivisions are markers for the

territorial boundaries of street gangs and therefore in some instances are lines of potential

conflict Some of those boundaries demarcate areas of Greater August Town that are

predominantly supportive of one or another political party but the boundaries do not always

hold political significance politics is but one element in the conflict geography of the area Like

many communities of the urban poor the Greater August Town area is easily mobilized

politically ndash a reality that is understood and at times exploited by street gangs who politicize

gang ldquowarsrdquo in their efforts to build alliances and to neutralize the police In fact the basic

principle of community mobilization in Jamaica is political patronage and clientelism Access to

resources (eg jobs housing education) for the poorest residents is often determined by the

local political party Thus according to Charles (2004 36)

supporters attach themselves to the political parties to get first preference in the

distribution of scarce resources and over time because they are unemployable they

become dependent on their political party for their economic survival These supporters

will kill anyone who threatens the support base of their political party because they

perceive it as a threat to their daily survival

As a consequence some political supporters invest heavily in the electoral contests and

provoke conflicts that affirm their loyalty to their party in order to secure material benefits

from it Political competition is one conflict fault line in what otherwise is a politically

heterogeneous community Specifically political support in Greater August Town is divided

between the Peoplesrsquo National Party (PNP) and the Jamaica Labor Party (JLP) The PNP receives

strong support from residents in August Town proper the upper region of Goldsmith Villa and

47

Bedward Gardens while the JLP is supported by those living in Hermitage and the lower region

of Goldsmith Villa (Charles 2004 ) The division between the political parties in the area as in

much of Jamaica was believed to be at the root of much of the communityrsquos violence

particularly between 1979 and 1993 Some gangs aligned with the PNP and others with the JLP

this often resulted in political boundaries overlapping with gang turf (Charles 2004)

Pre-truce Violence

Nationally the history of gang violence in the Greater August Town area first appeared as

political violence closely associated with the electoral cycle As in many other urban

communities the problem intensified mdash deeply affecting community life in the period just prior

to the national elections of 1980 mdash then continued cyclically until 1993 Gangs have since

harnessed this legacy in order to establish and maintain community support based on common

political affiliations They have used that support to nullify the efforts of law enforcement to

suppress their illicit activities (Harriott 2008)

The conflict profile of these gangs and of the communities in which they are nested has

changed over the decades From the beginning of their involvement in political violence and

territorial control a form of gerrymandering existed that manipulated the voting population

forcing certain individuals out of a particular constituency and preventing those who remained

in the community from voting for the opposing party This was linked both to the electoral cycle

and to the types of inter-gang conflicts that typically were associated with street gangs

(Figuerou Harriott and Satchell) More recently however much of the crime and violence

perpetrated by gangs has resulted from internal conflicts (eg status management disputes

over womengirls or money individual membersrsquo activities that could attract police pressure)

Internal conflict at times has led to gang fragmentation and new alliances that pull more parties

into the conflict escalating the homicide rate and increasing the sense of insecurity among the

general population (Levy 2012)

The most significant of these internal conflicts resulted from the killing of former Jungle 12

leader Neil Wright by members of his own gang Jungle 12 was the most influential gang in

Greater August Town Before his murder in order to increase the gangrsquos access to illicit

48

opportunities in Kingston Wright had been attempting to extend Jungle 12rsquos influence with a

system of alliances with other gangs and recruitment in Kingston (Harriott 2014) In short his

ambition was to transform Jungle 12 from a neighborhood street gang operating on the

outskirts of Kingston into to a dominant organized crime network that could reach into the

heart of the city In pursuit of this goal Wright recruited members from outside August Town

elevating them in the gang hierarchy above the locals This led to status-related conflicts and

resistance to Wrightrsquos leadership within the gang His murder precipitated a split of Jungle 12

into three factions two of them fled to other neighborhoods within August Town (Angola and

Vietnam) resulting in the formation of new alliances and a new conflict geography that

replaced the former political geography of conflict Wrightrsquos killing and the subsequent demise

of Jungle 12 as the dominant gang in Greater August Town altered the balance of power and

escalated inter-gang violence (Harriott 2014) The post-2005 phase of conflict was

characterized by power symmetry conflict intensification and the spread of conflict

throughout the entire geographic area of Greater August Town

Although their origins are unclear retaliatory killings and other violent incidents progressively

intensified between 2005 and 2008 The violence was episodic retaliations were most often

motivated by suspicions related to personal and geographic connections between warring

gangs As the violence escalated new alliances were formed to enhance power and dominance

which in turn increased the number of gangs and gang members involved in the violence

(Harriott 2014) This eventually attracted national attention and triggered community

mobilization for a gang truce

The Truce-making Process

The Greater August Town gang truce was preceded by frequent intense violence and public

outrage As noted above the violence had escalated in November 2005 when Jungle 12 leader

Neil Wright was killed The defection of a Jungle 12 member to Goldsmith Villa (Angola) caused

infighting within the gang and conflict between it and Goldsmith Villa Just a few months later

Wrightrsquos brother Steve and two others were injured during a turf battle (Martin-Wilkens 2006)

49

Thereafter violence began to occur at regular intervals until January 2007 when the Peace

Management Initiative (PMI) hosted a peace march in the community Two PNP politicians

urged the community to unite A PMI leader declared that the peace march was being held to

ldquodemonstrate to the public that Jungle 12 members are back together and that they want

peacerdquo (Thompson 2007 1)

Although hopeful some residents remained skeptical about the peace march perceiving the

action to be politically motivated In the absence of trustworthy information inter-group

conflicts tended to be interpreted through a politically partisan lens this created obstacles to

isolating the gangs building a consensus for peace and unifying community mobilization The

politically based narratives weakened the communityrsquos leverage for peace as well as the

exposure of the gangs to police action As one resident said ldquoThe election is coming up and

they want[ed] the people to vote for the PNP is one of the main reasons why they have to walk

todayrdquo (Thompson 2007 1) Those who shared such views stayed away from the peace march

Although that widely held myth was not factual it did serve to demoralize and demobilize one

part of the community A local UWI faculty member was articulate in his assessment of the

politics behind the march

August Town violence is not violence of organized crime which is based on drug

trafficking extortion or some other criminal enterprise [but] rather the violence in

August Town is essentially ldquotribalrdquo mdash the Peoplersquos National Party tribe versus the

Jamaica Labour Party tribe [which has been complicated by a splintering within the

PNP tribe] (2007 1)

Indeed it was not the violence of organized crime but neither was it political violence The

individual quoted above neglected to mention that the conflict was between those pro-PNP

splinter groups who were largely comprised of members of Jungle 12 Moreover their ldquopro-

PNP-nessrdquo was unrelated to the conflict there was no factional infighting within the local PNP

organization at that time

Nevertheless following the peace march the gang violence diminished Then in November

2007 a turf war erupted between two gangs from the Greater August Town neighborhoods of

50

Vietnam and River This time as the police stated the gang violence was less about politics and

more about dominance and turf Police were dispatched to perform directed patrols but

whenever they were not present the shootings continued (Mcleod 2007) In April 2008 the

community witnessed local gangs engaging in a five-hour-long street battle that left two killed

and three others wounded It ended only after the police deployed armored trucks The next

month another round of gang violence resulted in five others being killed including a one-year-

old child This resulted in the three members associated with the gang who committed the

homicides being killed in retaliation (Virtue 2008) The local community mobilized against the

violence increasingly cooperating with the police providing more information about the gangs

Subsequently gang members observed a decline in their influence within the community

During the early period characterized by low-intensity conflict the less influential gangs at

times used manipulation of the police as a tactic for suppressing the more influential gangs

This was largely done through strategic release of information Prior to 2005 when Jungle 12

was dominant its membersrsquo illicit activities were constantly reported to police by members of

other gangs as a means of compelling a compromise or settlement of conflicts In practice this

was done by ldquotrading casesrdquo Once a crime had been investigated by the police and suspects

had been charged an opportunity was created for the gangs and other parties to the conflict to

settle the matter by agreeing to drop their cases (typically by ceasing cooperation with police

investigators) This type of ldquoself-helprdquo served to end some of the retaliations but it rested upon

the manipulation of the police (Harriott 2014)

Later in an attempt to quell escalating inter-gang violence the police established buffer zones

between the warring gangs This action resulted in unintended consequences For example

when the police declared a buffer zone between August Town and Hermitage Hermitage took

advantage of the opportunity to attack Angola Some Angola residents accused the police of

turning a blind eye and creating an opportunity for Hermitage to attack their community

Although little reliable information exists about why the police made the deployment the way

that they did it is more likely that the police inadequately assessed the situation (ie mis-

assessed the pattern of alliances and the likely targets of attack) (Harriott 2014) In the areas

affected by this kind of increasing violence community members became angered and lost

51

confidence in the police The error resulted in some parties to the conflict receiving increased

support from their communities and in greater gang-community cohesion (Harriott 2014)

After a brief period the police identified this problem and began to disengage by no longer

providing a buffer between gang controlled areas which in turn allowed still more conflict to

occur between the gangs

As the violence escalated beyond their control police finally responded by applying their own

forms of pressure For example units under the direction of the JCF High Command would

make periodic raids in the community during which they would at times seize weapons and

make mass arrests (Sinclair 2005) However there were also moments when the local police

were very responsive improved their relations with the community and consequently gained

greater access to relevant information Two such moments occurred just prior to and again

immediately after the truce moments during which there was greater freedom of movement

and open collaboration between the community and the police (Harriott 2014)19

The Establishment of the Greater August Town Gang Truce

The Greater August Town gang truce was led by the Jamaican Peace Management Initiative

(PMI) The PMI is a government-funded initiative created for the purpose of working with gang

members to reduce violence Due to community mistrust of the police in 2002 the organization

was established as an alternative organizational mechanism for responding to gang violence

The PMI sought to bridge government and civil society efforts to mediate disputes between

gangs as well as to provide outreach to gang members (Bakrania 2013) While efforts to

institute a gang truce in Greater August Town were led by the PMI a number of other

stakeholders helped to facilitate the truce these included faculty at the University of the West

Indies (UWI) and representatives from the police the faith-based community and the August

Town Sports and Community Development Foundation (Jackson 2008 Levy 2009 also see

Appendix C) The gangs involved in the truce included those from August Town Hermitage

19 The quality of police-community relations largely depended on the style of the local station commander however regardless of the external environment

52

Goldsmith Villa Bedward Gardens and African Gardens Because of its formality as well as its

perceived effectiveness the truce signed on June 24 2008 was regarded by many as the first

of its kind in Jamaica (Levy 2009)

Truce negotiations began early in June 2008 and lasted for about three weeks The gangs

sought to leverage their violence-making capabilities and demanded payment for peace They

asked the third-party negotiators for money ldquoworkrdquo and start-up funds for proposed micro-

businesses (Wilson 2014) Those demands were rejected by the negotiators on the grounds that

the third-party institutions would not buy a peace that was intended to save the lives of those

who were making the demands Moreover if peace was to be purchased then gang conflict

could be used continuously to extract money and other benefits from negotiators The third-

party actors made some demands of their own In some quarters of the community and society

the surrender of guns was viewed as a litmus test of the sincerity of the gangs Consistent with

this the negotiators suggested that the parties to the conflict symbolically hand over one gun

each that suggestion was immediately rejected by the gang leaders These kinds of demands

from the various parties ceased after a time as they all agreed that the truce was to stand on

its own merits (Harriott 2014)

As the truce began to be committed to paper a number of stakeholders expressed concern that

their greatest risk in participating could be the potential for Jungle 12 factions to use the peace

agreement as they had in the past as a tactic to persuade their enemies to let their guard

down Others however recognized that Jungle 12 had now been weakened and that a formal

public peace agreement would be beneficial to the gang and therefore this time would be

different (Harriott 2014)

The gangs held fast to their claim that their weapons were needed for their own protection

because the police were ineffective in responding to violence in their communities (Jamaican

Gleaner 2014) It became a precondition of the truce that the gangs would not be required to

turn in their guns and other weapons (Jackson 2008) The truce agreement did specify

however that ldquoall persons are allowed to move freely across all boundaries regardless of

reputation or affiliation No gun salute or any other shooting is to take place in the community

53

for a period of at least five yearsrdquo (2008 also see Appendix C) The truce agreement and its

conditions were prescribed in a document that was finally signed by all of the major

stakeholders including the gangs (see Appendix C)

Throughout the negotiations each of the gang leaders had attempted numerous times to use

the truce as an opportunity to bargain for money jobs and business support grants Such

demands consistently were rejected by the third-party actors Nonetheless both prior to and

after the truce some efforts were made to create better opportunities for young people

residing in the community UWI for example provided a homework supervision program to

encourage students to further their education and it developed a community-building initiative

to help improve schools and enhance sports programming (Levy 2009) Such programs were

conducted as part of UWIrsquos Township Project in August Town which invested significant

resources in developing the residentsrsquo job-related capabilities and collective self-efficacy

The Greater August Town gang truce was noteworthy for two reasons First the gang truce

received substantial press attention The media were invited to witness the ldquosigningrdquo of the

truce by the gang leaders in the presence of a JCF Deputy Commissioner of Police a PMI board

member and the UWI Principal and two professors of its faculty Second the truce was widely

credited with decreasing violence in Greater August Town and it served as an exemplar to

other communities seeking to replicate its success (Virtue 2008) A number of reports

manuscripts and newspaper articles proclaimed the truce to be a success Bakrania (2013 10)

for example reported that ldquoPMI has been credited with stopping gang wars in August Town

rdquo Levy (2009 94) remarked that the ldquomost interesting outcome of PMI efforts to date was the

Peace Agreement reached in August Town in late 2008rdquo Likewise a government report noted

that ldquothe peace treaty was a pivotal achievement in August Town that has significant potential

for wider application Crime levels dropped markedly in August Town after the signing of the

peace agreement in June 2008rdquo (McLean and Blake-Lobban 2009 78) To this day August Town

celebrates the signing of the truce with an annual celebration with food and music

(Cunningham 2011)

54

Methods

Our evaluation relied on a pre-testpost-test quasi-experimental group design Our

methodology examines the Greater August Town community which is comprised of three

contiguous towns where the gang truce took place (the target area) and the balance of Jamaica

which is comprised of 178 communities (comparison areas) As seen in Exhibit 12 the average

number of residents living in each of the three communities in the target area was not

significantly different than that for the rest of Jamaica about 7776 residents lived in each of

the Greater August Town communities compared with 6468 in the other communities

Likewise communities of Greater August Town were about as densely populated as other

communities (2960 per square kilometer versus 2647 per square kilometer) and the age

range of residents was similar as well However Greater August Town (a) had a significantly

higher proportion of its residents living in poverty (196 vs 158) (b) consumed fewer

resources than other communities and (c) reported significantly more homicides than other

communities prior to the truce (see Exhibit 12)

55

Exhibit 12 Descriptive characteristics of Greater August Town and balance of Jamaica (2007-2011)

Comparison

Area Greater August

Town All areas

Population (mean) 6468 777633 648994

(sd) 720482 353731 715621

Population density (mean) 264719 296033 265238

(sd) 271023 285501 270465

Percent in poverty 1577 1957 1583

(sd) 1036 106 1029

Consumption 15737890 1106939 1566048

(sd) 10713020 205336 1064021

residents under 15 yrs old 2369 2494 2371

(sd) 487 115 484

residents 15-65 yrs old 6840 6901 6841

(sd) 423 29 419

Murder per month (mean) 674 857 677

sd 1928 1409 1920

Total murders 10068 180 10248

n 178 3 181

plt=05

Measures

Two distinct data sets were merged to measure the impact of the Greater August Town truce

First data from the 2011 decennial census provided community-level measures of the social

and economic characteristics of the 181 communities in Jamaica Described in detail below the

community-level data used in the study included population population density gender age

poverty and consumption20 These data were obtained directly from the Statistical Institute of

Jamaica

Second police homicide data from the years 2007 through 2011 were used to construct the

studyrsquos community-level measure of homicide The homicide data were aggregated by month

20 Consumption is an alternative measure of poverty in Jamaica which measures the consumption of food and non-food items

56

and appended to the community-level data The final (merged) data set included 10248

homicides over the 60-month study period These data were obtained from the Jamaica

Constabulary Force (JCF)

The dependent variable examined in the study was constructed from official police homicide

data Once again the homicide data represented the number of officially recognized homicides

in Greater August Town and each of the remaining communities in Jamaica We examined

change by comparing the homicide data 18 months prior to the truce with the homicide data 42

months following the truce More specifically we examined whether there was a change in the

number of homicides in the 30 days following the truce (month 1) as well as whether the truce

had an impact every three months thereafter (ie months 2-5 6-8 9-11 12-14 15-42) and

whether any changes in homicide coincided with changes in homicide in the balance of

observation areas The frequency distribution of our dependent variable is presented in Exhibit

13 It shows that prior to the truce the target area on average experienced significantly more

homicides (1495) than did the comparison areas (920)

57

Exhibit 13 Distribution of homicides in the target and comparison areas

Comparison

Area Target

Area Total

Pre-truce period Mean 920 1495 932

SD 2469 1966 2461

N 241400 5100 246500

Month 1 of truce Mean 741 286 733

SD 1785 496 1772

N 17800 300 18100

Months 2 thru 5 of truce Mean 647 905 652

SD 1821 1249 1812

N 71200 1200 72400

Months 6 thru 8 of truce Mean 577 1236 588

SD 1690 1074 1683

N 53400 900 54300

Months 9 through 11 of truce Mean 718 333 711

SD 2034 733 2019

N 53400 900 54300

Months 12 through 14 of the truce Mean 687 095 678

SD 1519 286 1509

N 53400 900 54300

Months 15 thru 42 of the truce Mean 564 589 564

SD 1683 1042 1674

N 516200 8700 524900

Total Mean 674 857 677

SD 1928 1409 1920

N 1006800 18000 1024800

An illustration of the trends in homicide prior to and following the gang truce are shown in

Exhibit 14 It shows that 30 days following the truce homicides fell in the target and

comparison areas then increased and decreased several times with a general downward slope

in violence over time

58

Exhibit 14 Monthly number of homicides pre-post truce in the target and comparison areas

We also used a number of measures to control for community-level structure from the 2011

decennial census These community-level data included the communityrsquos population

population density (per square kilometer) and community level of consumption Additionally

the census data included measures of the percentage of the population that was female under

15 years old 15 and 65 years old and 65 years old and older as well as a measure of the

percentage of the population living in poverty Principal components analysis was used to

reduce some of these data into a summary measure

Exhibit 15 shows the results of the component loadings One component was extracted that we

designated as socio-economic status (SES) which exhibited high loadings for percent living in

poverty percent under 15 years old percent 15 to 65 years old and consumption Excluded

from the principal components analysis were population and population density Population

was used as our exposure variable and population density was logged to address skewness in

these data

Exhibit 15 Factor loadings from principal components factor analysis

Loading Poverty 78 Consumption -76 under 15 years old 92 between 15 and 65 years old -80

59

Analytic Strategy

In order to test whether the truce had an impact on homicides in the target area andor

whether displacement had occurred in the balance of the study area several analytic

techniques were employed Most of the methods employed the use of the homicide rate as the

dependent variable We explored the data in this way to provide the maximum statistical

power to detect an effect As a check on these methods we also employed a generalized model

to compensate for the non-normality of our outcomes

First focusing only in the target area we performed a simple t-test comparing the homicide

rates before and after the truce (the unit of analysis was a month) however this technique had

limitations the most severe of which was that even if the test were significant it would be

difficult to determine whether the difference was due to the gang truce or a natural change

over time in the outcome Second to address this limitation time series models were

employed whereby the homicide rate for the target area was modeled as a function of time

with truce period indicators included to measure the effect of the truce net of the temporal

trends These models were estimated with ARIMA techniques with a one-month lag auto-

correlated error Third we examined the homicide rate for each town using a panel time series

model In this model the temporal trend for each town was examined with indicators for target

areas and truce periods included The main effects for the truce periods measured the effect of

the truce in the target areas and the moderators of the truce period in the comparison areas

measured displacement effects Finally because the dependent variable coded is not normally

distributed across months we used a negative binomial time series model to estimate the

number of homicides with the population covariate serving as an exposure variable

Findings

The first set of results examines only the target area The first test was a simple t-test

comparing the mean homicide rates before and after the truce periods The result was a mean

difference in the homicide rate of -890 per 100000 with a significant t-statistic of 370 While

60

this result is statistically significant we caution that it may or may not reflect an impact of the

truce To further examine the truce effect in the target area we performed ARIMA regressions

The first model did not include an effect of the temporal trend In this model we again find that

by month 15 the murder rate decreased by about -89 per 100000 (Exhibit 16)

Exhibit 16 Results of basic ARIMA model

Next we employed the ARIMA model again but included a variable (date) to control for the

temporal trends in the data Exhibit 17 shows that when we controlled for temporal trends the

impact of the truce we observed was no longer significant This result indicates that it was not

the truce per se that caused the decline in homicides but instead the decline in homicides was

part of a larger (local and nationwide) decline in homicides

Note The test of the variance against zero is one sided and the two-sided sigma 7877339 5907266 1334 0000 6719536 9035142 L1 1720758 1242218 139 0166 -0713944 415546 ar ARMA _cons 1538436 1904285 808 0000 1165203 1911669 t_15 -8888017 2969383 -299 0003 -147079 -3068134 t_12 -1507749 1820406 -083 0408 -507568 2060181 t_9 -132475 9189969 -144 0149 -3125951 4764507 t_6 -2861008 6335492 -045 0652 -1527835 9556328 t_2 -5166363 4498142 -115 0251 -1398256 3649834 t_1 -1111771 4463623 -025 0803 -9860311 7636769murders_rate murders_rate Coef Std Err z Pgt|z| [95 Conf Interval] OPG

Log likelihood = -2089969 Prob gt chi2 = 00397 Wald chi2(7) = 1472Sample 564 - 623 Number of obs = 60

ARIMA regression

61

Exhibit 17 Results of ARIMA model with control of temporal trends

We next estimated the possible displacement effects of the truce Exhibit 18 presents the

results of these models Examination of the main effects of the truce period does not indicate

any effects and looking at the truceComparison interaction effects we also do not find any

displacement effects Note that these models also controlled for the socio-demographic

characteristics of each community

Note The test of the variance against zero is one sided and the two-sided sigma 7775398 6622814 1174 0000 647735 9073446 L1 1270618 1253989 101 0311 -1187156 3728392 ar ARMA _cons 1480274 1197324 124 0216 -8664378 3826986 date -2318963 2091772 -111 0268 -6418761 1780835 t_15 -3519687 7694691 -005 0964 -1543329 1472935 t_12 -970747 2228007 -044 0663 -5337561 3396067 t_9 -8844212 9402792 -094 0347 -2727335 9584921 t_6 2995697 6576644 005 0964 -1259042 1318956 t_2 -2353533 4923798 -048 0633 -12004 7296933 t_1 -9194625 5462116 -017 0866 -1162501 9786088murders_rate murders_rate Coef Std Err z Pgt|z| [95 Conf Interval] OPG

Log likelihood = -2082031 Prob gt chi2 = 00710 Wald chi2(8) = 1444Sample 564 - 623 Number of obs = 60

ARIMA regression

62

Exhibit 18 Results of Panel (Town) time series model with control of temporal trends

Last we used a random effects negative binomial regression that predicted the homicide rate

with population as an exposure variable and controlled for the socio-demographic

characteristics of each community The results are presented in Exhibit 19 The analysis showed

that time had a negative effect indicating that homicides were decreasing in all areas over the

study period The main effects of the truce (truce = 1 2 ) represented the effects of the

truce in the targeted area and did not show a significant effect for any period following the

gang truce However we did find that the homicide rate significantly increased in the

_cons 6310151 3518738 179 0073 -5864483 1320675lnpopulationdensity 121961 1709831 713 0000 8844891 1554731 ses 1245334 2428629 513 0000 7693312 1721336 date -1276623 0345659 -369 0000 -1954101 -0599144 15Balance 5392222 4128863 131 0192 -2700202 1348464 12Balance 1318516 7903159 167 0095 -2304746 2867507 9Balance 1042658 7925876 132 0188 -5107854 2596101 6Balance -1703928 7919006 -022 0830 -1722489 1381704 2Balance 3498997 7175155 049 0626 -1056405 1756204 1Balance 1056874 1104696 096 0339 -1108291 3222039 truceaugust Balance -4757021 3276618 -145 0147 -1117907 1665032 august 15 -4269255 4284086 -100 0319 -1266591 41274 12 -1262496 7871907 -160 0109 -2805361 2803697 9 -1014276 7881965 -129 0198 -2559113 5305603 6 6563019 786731 008 0934 -1476334 1607595 2 -5034368 7123478 -071 0480 -1899613 8927392 1 -1215158 1095699 -111 0267 -3362688 9323715 truce murders_rate Coef Std Err z Pgt|z| [95 Conf Interval]

Prob gt chi2 = 00000 Wald chi2(16) = 12359 max = 60 avg = 5661878Estimated coefficients = 17 Obs per group min = 48Estimated autocorrelations = 1 Number of groups = 181Estimated covariances = 1 Number of obs = 10248

Correlation common AR(1) coefficient for all panels (02045)Panels homoskedasticCoefficients generalized least squares

Cross-sectional time-series FGLS regression

63

comparison areas in months 12 through 14 following the truce In particular we found a 29

percent increase in the homicide rate in the comparison communities for that period (exp (-

1797 + 2048) = 1285 p-value = 004) Since this effect is only significant at the 005 level

however and given the number of analyses used to examine the data it is possible that we

found this effect by chance alone

64

Exhibit 19 Random Effects Negative Binomial

65

Given these caveats we visualized this model with the following set of marginal predictions as

observed in Exhibit 20 We saw that the targeted area (as illustrated in red) experienced an

immediate decrease in homicide which coincided with an increase in homicides in the balance

of the study area However the target area quickly returned to ldquonormalrdquo and homicides in the

comparison area decreased again During months 9 through 11 following the truce there was a

reduction in homicides in the target area with an associated increase in the comparison area It

is important to point out that the confidence intervals are large and we cannot yield concrete

conclusions from these results However it appears that the truce might have had a temporary

short-lived displacement effect decreasing homicides in the target area but increasing

homicides in the comparison area

Exhibit 20 Predicted change in homicides in the target and comparison areas

Conclusions

From 2000 through 2009 Jamaica experienced a substantial number of homicides many of

which were attributed to gangs in one form or another Traditional law enforcement responses

were repeatedly implemented but until 2010 those had little effect Some policymakers in

Jamaica as well as in other nations throughout the Caribbean and Central America have

recently been experimenting with novel approaches to reducing gang-related violence notably

the implementation of gang truces In Jamaica at least eight gang truces reportedly have been

66

negotiated since 2001 (Levy 2009) The Greater August Town gang truce was thought to have

been one of the more successful and it has served as a model for other communities to use

(2009) Our purpose here has been to identify the actors involved in the negotiations of that

truce the negotiation goals and the implementation methods used and then to examine

empirically the impact of that truce on homicide rates in the targeted community

The 2008 gang truce in August Town was a response to violence that arose when the leader of

one gang was killed creating a power vacuum that other gangs saw as an opportunity to

increase their influence in the community Concomitantly the community as well as the gangs

feared that an absence of formal social control would result in further violence The police

reacted unevenly At some times they engaged in appropriate but heightened levels of

preventive patrol while at other times they purposely provided little or no protection on

occasion they used aggressive tactics that further isolated them from the community The end

result was that there was neither stability nor predictability in the police response and

therefore little trust in the police to address the problem

As the violence further escalated the community mobilized The Jamaican Peace Management

Initiative faculty members from the University of the West Indies (UWI) the Jamaican

Constabulary Force (JFC) and other community-oriented groups joined forces seeking to

reduce the increasing number of homicides by brokering a truce between the gangs Over the

three-week negotiation period the negotiators and the gangs sought terms from one another

The gangs wanted payment ldquoworkrdquo and funds for micro-business development to end the

violence The third-party stakeholders wanted the gangs to disarm actually or symbolically

Neither the gangs nor the stakeholders had substantial leverage nor did they have much to

offer one another in terms of incentives In the end however a truce was agreed upon and all

of the gangsrsquo leaders and several key community stakeholders signed it at a public ceremony

with the media in attendance

At first glance our impact findings appeared to show that the gang truce was an effective

mechanism for reducing violence Bivariate analyses showed a significant decline in homicides

after the truce was implemented This explained the work previously published by

67

policymakers researchers and news reporters Upon further examination of the data however

comparing change in the target and comparison areas and accounting for temporal trends we

found that the decline in homicide was part of a larger nationwide decline in violence and that

the gang truce was not responsible for the decline The only significant effect that we

uncovered was that possibly the homicides were displaced outside the target area for a brief

period of time but then returned to normal

Any one of a number of explanations might be offered for the strategyrsquos lack of effectiveness It

might be that the Jamaican gang leaders at least those in Greater August Town did not have

the organizational capacity to change gang member behavior Much prior research suggests

that in general gangs have limited organizational structure and little formal leadership This

might suggest that gangs do not possess the necessary capacity to regulate their membersrsquo

violence That said gangs in Jamaica including in Greater August Town have been found to be

fairly organizationally sophisticated and to possess strong leadership

In fact in a small number of Jamaican communities gangs have been found to be highly

organized with individual gang leaders being referred to as dons and community leaders The

gang leader in such a community is often found to have substantial control over members and

residents as these communities often turn to the don rather than the police for justice The

don will hold court and punish those who commit crime Punishment can include beatings and

torture as well as execution (Morgensen 2004) Although this level of organizational structure

and sophistication is found only in a small number of Jamaican communities generally the

gangs in Jamaica are believed to have some organizational capacity or at least enough to

reduce violence in communities

Our findings however indicated that prior to 2005 and the death of Neil Wright perhaps only

Jungle 12 could approximate that capacity to discipline members and enforce a truce After the

gangrsquos fragmentation in 2005 Jungle 12 lost much of its organizational capability and

enforcement of the truce was therefore difficult The truce negotiators sought to address the

enforcement issue by proposing a peace council that would involve all parties The proposal

was approved by all key stakeholders still some gang leaders demanded cash payments as a

68

condition for attending council meetings Peace was consistently seen by then as a bargaining

tool rather than as an honest attempt to establish and maintain peace In the end members of

only two gangs were attending the meetings21 and the council soon dissolved

In an effort to replicate the council function UWI sponsored one of the most respected

negotiators a community activist to become a one-person monitoring and intervention

specialist or a ldquoviolence interrupterrdquo His job was to ensure that truce violations did not lead to

a return of the gang wars mdash and there were many violations of the truce For example there

were instances of gang members crossing boundaries and entering the turf of another gang

armed although not initiating conflict behavior that was interpreted by the opposing gangs as

preparation for the next round of ldquowarrdquo or as laying a foundation for a surprise attack that

would exploit the truce for this purpose In the absence of the council these matters were

reported to the violence interrupter who tried to resolve the problems in consultation with the

various gang leaders Often the gang leaders were unresponsive or incapable and therefore the

threatening practices and violence continued Ultimately there were no rules or bodies or

persons who could regulate the violence and there were never any reference points for

compliance The formal truce agreement was an attempt to negotiate and impose such rules

via a collective pressure that would include third parties but it was unsuccessful in doing so

The potential for re-engineering norms related to conflict thus was not realized

Another explanation for the failure of the gang truce might be that it was more a vehicle for

rhetoric rather than for reality The gang leaders insisted that they would sign the truce

agreement only if it were ratified in public with the presence of the media (Jackson 2008 Levy

2009) The leaders might have viewed the process in and of itself as a means of increasing their

reputation and influence within the community and in policymaking circles (and to reduce

mutual distrust) In signing the truce gang leaders publicly pledged to reduce their involvement

in violence thereby calming local residentsrsquo fears They also made public efforts to increase

resources for their communities perhaps in an attempt to portray themselves as ldquoprovidersrdquo to

the community In fact the truce did provide gang leaders with an opportunity to be seen in

21 Interestingly the Jungle 12 factions did not attend any of the peace council meetings

69

public collaborating with important community stakeholders The imagery of the public signing

was of the government (via the PMI) and others approaching the gang to ask them to use their

means of informal social control in the community to reduce violence mdash to accomplish

something that the government could not do on its own As a consequence the process may

have been perceived by gang leaders as a victory because it enhanced the gangsrsquo reputation

with both the government and community

Alternatively from the start the gangs might not have been fully invested in the gang truce

One of the major criticisms of the Greater August Town gang truce was that gangs were not

required to give up their firearms although some believed that this was an unrealistic request

their demand and the demand of many that all guns be turned in immediately was

quite unrealistic given the decades of ingrained gun culture and the continued inability

of the security forces to guarantee protection for any corner against armed rivals It was

obvious to most observers that that kind of situation could not be ended overnight and

that this was a reasonable first step in the process (Levy 2009 63)

The gangs feared that if they were to disarm themselves they would be vulnerable to other

gangs and unable to protect themselves a concern that appears not to have been addressed by

mediators Indeed at times some elements within the community felt somewhat dependent on

the gangs to maintain security If the gangs would have been disarmed and there were no

near-term alternative prospects for any form of social control both the gang and the

community might have faced additional violence as has been observed in the past In the end

the gang truce only called for a reduction in gang violence and did not provide any solutions to

address the larger problems between the gangs nor did it provide the gangs with any tangible

benefits for abiding by the truce

70

Case Study C Gang Trucemdash The Honduran Experience

Introduction

Violence in Honduras is at epidemic levels increasing almost 44 percent over the past five

years In 2012 there were 7172 homicides in Honduras or about 86 homicides per 100000

population (Instituto Universitario de Democracia Paz y Seguridad 2013) making it the most

violent nation in the world (United Nations 2013) Likewise Hondurasrsquo second largest city San

Pedro Sula has the highest municipal level homicide rate in the world with 1290 homicides

(Instituto Universitario de Democracia Paz y Seguridad 2013) or about 174 per 100000

population (United Nations 2013) In comparison the average homicide rate across the globe is

about 62 per 100000 and the average homicide rate in Central America is about 27 per

100000 (United Nations 2013)22

Much of the discussion about the causes of Hondurasrsquo high homicide rate has focused on its

relationship with international drug trafficking routes gangs and conflict between crime

groups and the government and government instability Estimates of gang involvement very

widely but some have suggested that there are between 12000 (Seelke 2012) to 36000

(Ratcliffe et al 2014) gang memberrsquos in Honduras who typically belong to one of two gangs

MS-13 and 18th Street These gangs are said to be less organized than their counter parts in El

Salvador but are said to be just as involved in extortion and intimidation and perhaps more

involved in drug trafficking because of their stronger linkages with Mexican drug cartels

(Wilkinson 2013)

22 An unusual characteristic of the homicide problem in Honduras is the age of victims Typically in the Western Hemisphere homicide victims are aged 15 to 29 In Honduras however those 30 to 44 have the highest rate of violent victimization For example 1 out of 280 males 30 to 44 years old are the victim of homicide compared to 1 out of 360 males 15 to 29 years old (United Nations 2013) These findings by themselves are suggestive of a chronic gang problem (Spergel 1995) Honduras also stands out in the Western Hemisphere in the proportion of its homicides that involve a firearm In 2012 about 84 percent of the homicides involved a firearm (11) The proportion of homicides that involve a firearm appears to be increasing as well In 2008 79 of homicides involved a firearm compared to 81 in 2009 83 in 2010 (United Nations 2013)

71

Over the past decade the nation has responded with ldquoMano Durardquo (ie iron fist or heavy hand)

The new legislation provided the police with more authority to stop search and detain gang

members The new legislation also permitted the courts to sentence gang members to prison

for 12 years for simply being a member of a gang and allowed the courts to sentence

individuals to even longer prison terms for gang related incidents Concomitantly the military

joined the effort to fight gangs by patrolling neighborhoods along side the police While the

public and media strongly supported the shift in national policy toward Mano Dura much of

the evidence suggests that these legislative and policy changes were not effective as the

number of homicides continued to escalate Some suggest that its lack of success was because

gang members who were arrested were released due to of lack of evidence or those who went

to prison if they were not a gang member before entering prison joined a gang Other critics

point out that the heavy handed approach by the government led to loss in the rule of law as

vigilantes engaged in extra-judicial violence against gang members (Seelke 2012) Still others

said that the legislation and policies never really had a chance of working because of the

general lack of effectiveness of the police and courts and the wide spread corruption

throughout the criminal justice system (Zilberg 2011)

As a consequence of the above policymakers and citizens voiced optimism about the possibility

of a truce between gangs after initial results in El Salvador suggested the strategy might be

effective (Villiers-Negroponte 2013) Honduran church leaders and the Organization of

American States (OAS) began to develop a strategy to implement a similar type of truce in

Honduras and the President offered his personal support in their efforts (Arce 2013) In this

case study we examine the processes that lead to the Honduras gang truce and the nationwide

impact of the truce on homicides In the below section we discuss the major stakeholders who

participated in the truce processes leading up to the truce and the establishment of the truce

Key stakeholders

The primary facilitator for the truce process in Honduras was Archbishop Roacutemulo Emiliani who

had earlier served as the Assistant Bishop of the Dioceses of San Pedro Sula and who received

72

support from the Catholic Church to pursue the truce (Bosworth 2013) Prior to the

negotiations he was well known for his work which attempted to establish peace between the

gangs and his advocacy for prison reform and social reintegration programs for gang members

(The Daily Herald 2013) From the onset Monsignor Emiliani proceeded cautiously to ensure

reasonable expectations among the public and policymakers He maintained publically that ldquohe

didnacutet want to be a salesman of false promises about what was going to occur in the future the

things that they do are unpredictable but we expect to have a declaration of reconciliation

principles with societyrdquo (El Mundo 2013) Additionally he wanted to set reasonable

expectations because he knew that it would be a ldquoslow painful and draining processrdquo (El Nuevo

Siglo 2013) and that ldquoWhat is coming is difficult It is not easy It is complicatedrdquo (Castillo

2013)

As in El Salvador the Organization of American States (OAS) played a major role in facilitating

the peace process alongside Monsignor Emiliani Adam Blackwell served as the Secretary of

Multidimensional Security for the OAS and represented Canada on the Honduras Security

Reform Commission (Willcocks 2014) His participation in the mediation process was requested

by Honduran gang members who were in prison They requested that the OAS help broker a

peace agreement with the Honduran government and to help identify resources that would

assist gang members to obtain legitimate jobs (Associated Press 2013) The OAS together with

the Catholic Church served as a ldquobridgerdquo between the executive branch of the government and

the two gangs Additionally two of the mediators (ie Salvadoran Army officer and Police

Chaplain Monsigor Colindres and former Salvadorian congressman Mijango) who helped broker

the truce in El Salvador provided additional support to Monsignor Emiliani and Secretary

Blackwell They traveled to Honduras to present their experiences with the gang truce in El

Salvador and to convey that a gang truce is a promising and legitimate strategy for addressing

gang violence (Associated Press 2013)

Gang leaders of the two primary gangs in Honduras (MS 13 and 18th Street) also participated

extensively in the negotiation process It was stated that they had become weary of the violent

conflict and understood that a truce would be beneficial to the Honduran people (Servellon

73

2013) From the beginning however a number of the critics of the truce argued that Honduran

gangs did not have the capacity to control street level violence They characterized the

Honduran gangs as having less organizational leadership (Bossworth 2013) less control over

turf (LatinNews Daily Report 2013) and being more organizationally fractured (Dudley 2013)

than MS 13 in El Salvador

At the time that discussions about the possibility of a gang truce began President Porfirio Lobo

Sosa was publically supportive of the Catholic Church and OAS negotiating with the gangs

Media reports quoted the President saying I am ldquoprepared to do what ever is necessaryrdquo to

support the mediators (Phillips 2013) ldquoWe have to look for anything thatrsquos an alternative to

violencehellipOn the part of the government we are open to any process that can lower violencerdquo

(Associated Press 2013) and that he had given ldquohis blessing to Emillanirsquos efforts to broker peace

between the gangshelliprdquo (The Daily Herald 2013) However in November 2013 after a general

election the new president Juan Orlando Hernaacutendez through his recently appointed Vice

Minister of Security declared that the government would no longer support the truce process

with the gangs (El Heraldo 2014) Since then the Government of Honduras has not mentioned

the peace process that was initiated in May 2013

Truce making process

It is important to note that prior to the announcement of the gang truce a number of key

stakeholders were somewhat skeptical about its possibility On the one hand some suggested

that a gang truce had been attempted in the past with no success For example one

stakeholder commented to an international media outlet that ldquoEveryone here agrees itrsquos a

positive step forward but people are cautiously optimistic because in 2005 these two gangs

had another peace treaty with each other Now that treaty was very tentative it only lasted

less than two monthsrdquo (Al Jazeera 28 May 2013) On the other hand as noted above other

stakeholders believed that the local gangs did not have enough organizational leadership to

change the behavior of gang members to reduce violence (Bosworth 2013) They argued that

even if gang leaders wanted a gang truce there was no way of enforcing it on the streets

74

Several months prior to the announcement of the gang truce Carlos Mojica Lechuga an 18th

Street Salvadorian gang leader publically stated that representatives of MS13 and 18th Street

in Honduras spoke with several gang truce key stakeholders in El Salvador for the purpose of

replicating the truce in Honduras Tellingly in reflection of the visit Mojica noted that the

advantage of a gang truce is that it formally recognizes Honduran MS13 and 18th Street leaders

as important political persons within the nation He also noted that Honduran gang leaders

have historically been treated poorly and that a gang truce holds the potential for

demonstrating the political power of each of the Honduran gangs (Villiers Negroponte 2013)

The negotiators used different language to describe the early days of the truce Specifically

they mentioned that there had been a consultative process with the gang leaderships so they

were in a process like lighting23 Prior to the truce media sources mentioned that the gang

leaders were offering to stop violence and to not recruit more youth into the gangs24 The gang

leaders also spoke about the ldquopersecutionrdquo they and their family members had suffered during

the previous years highlighting that they had been prohibited social opportunities offered to

others in society (Arce 2013)

Leading up to the negotiations the leadership of both gangs expressed their interest in three

goals 1) lowering violence and crime 2) reconciliation with God society and the government

and 3) helping to improve the social conditions of their communities Although there was not

written documentation on the exact agreement between the parties one MS-13 leader

affirmed that the pact would include all violence (El Comercio 2013) However when talking

about sensitive topics such as extortions which is one of the main sources of income for the

gangs the gang leader said that ldquowould be taken up at a later daterdquo (ABC Internacional 2013)

Leaders of 18th street made similar general statements about ending violence but they were

more specific about their demands One of the 18th Street leaders stated that ldquowhat we want is

23 El Universal 2013 - httpwwweluniversalcominternacional130531obispo-hondureno-descarta-tregua-entre-las-pandillas 24 (Garcia 2013 -httpwwwlaprensahncspmediapoolsitesLaPrensaHondurasSanPedroSulastorycspcid=338546ampsid=276ampfid=98-)

75

to have a dialog with any commission appointed by President Porfirio Lobo and we are sure

that the situation in Honduras will begin to changerdquo (El Nuevo Siglo 2013)

Establishing the gang truce

On May 28th 2013 with public declarations from leaders of both gangs the gang truce was

announced From the beginning of the process the role of the government in the truce was

unclear (La Prensa 2013) Likewise there was little discussion about the exact nature of the

agreement the terms in which gang members would abide and any benefits that would be

made available to those who participated in the truce For example as one stakeholder

indicated the government never decisively considered viable proposals to give the members of

the gangs any opportunities It is important to note that none of the parties signed a formal

commitment and neither MS-13 nor 18th Street signed any type of ceasefire agreement The

gang leaders were in separate locations and were never in direct contact with each other during

the announcement That is both gangs seemingly agreed to the gang truce without ever

talking to each other The ldquopeace processrdquo was publicly announced on May 28 2013 through

ldquojoint but separaterdquo declarations made by the leaders from both gangs imprisoned in the San

Pedro Sula prison (National Penitentiary SPS) (The Daily Herald 2013) The national and

international media widely covered the declarations

The MS-13 leaders said they would not commit any more homicides or any other types of

crimes They ensured that this was an ldquoimmediaterdquo order and would be effective throughout

the country They emphasized ldquoall of the boys know what they have to do starting todayrdquo

(Pachico 2013) The leadership of 18th Street declared that they would stop violence and other

criminal activities but also indicated that the government would have to ldquolisten to themrdquo

Little research has examined whether the gang truce in Honduras ever impacted violence in the

nation Instead anecdotes have been used to portray its effectiveness One facilitator for

example indicated ldquoin Honduras the dialog with the gangs has been positive however the sad

thing in Honduras is that the two main gangs have not accepted a truce between them they

just haven`t accepted it as yetrdquo (La Prensa 2013) despite the fact that ndash in their own words ndash

76

ldquothey do want to hold a dialog with society with the government and with the policerdquo

Conversely gang leaders declared that the truce had been effective For example a leader of

18th Street noted that ldquohellipIt has already done its part telling members in the areas the gang

controls to stop the violence and crimehellip[estimating] crime had already dropped 80 percent in

those areasrdquo (Associated Press June 17 2013) Similarly a member of MS13 estimated that

violence in MS13 controlled areas declined by 45 percent (Associated Press June 17 2013) As

a symbolic gesture of the impact of the truce MS13 leaders also noted that as a gesture of good

will they made and delivered 60 beds for a nursery home in San Pedro Sula (Associated Press

June 17 2013)

Methods

For the present case study we used a pre-test post-test longitudinal quasi experimental design

Data from the 2001 Honduran Census was obtained from the National Institute of Statistics

(INE) These data provided municipal level measures of number of residents population density

per kilometer percent of population who moved in from another municipality ethnicity

percent urban number of residents immigrated to the United States percent female headed

households percent unemployed age composition income percent of households rented and

education level In addition population projections for the years 2005 2010 and 2014 were

also obtained from the National Institute of Statistics (INE) The population levels for the total

population as well as the percentage of residents in a municipality that rural and percentage of

residents who are female were linearly interpolated for the intervening years Examination of

the observed levels indicated that growth was linear overall and so we feel confident that our

linear interpolations are good approximations Second we used municipal level homicide data

by month and year for the period May 2012 through July 2014 These data were provided by

the Honduran National Police through the Honduran US Embassy Both datasets were

merged for the present analysis

Measures

77

The dependent variable for the Honduran case study is the monthly homicide rate which was

calculated by dividing the number of homicides in each municipality by its population and

multiplying this figure by 100000 We examined change by comparing the homicide data 13

months prior to the gang truce to the homicide data 14 months following the truce in each of

the nationrsquos 298 municipalities As presented in Exhibit 21 there were a total of 7910 homicides

over the study period with each municipality averaging 183 homicides (sd=837) A trend

analysis showing the monthly number of homicides on a national level prior to and following

the gang truce is presented in Exhibit 22 It shows that nationally the homicide rate gradually

declined over the study period

Exhibit 21 Summary Statistics

Pre-Truce (n=3809) Post-Truce (n=4101) Total (n=7910)

Mean SD Mean SD Mean SD

Homicide 200 900 166 773 183 837

Population 2874492 8674659 2939557 8870267 2908226 8776124

Homicide rate 521 953 413 814 465 885

Density 9926 10878 9928 10878 9927 10877

Born in the same municipality 8211 1286 8211 1286 8211 1286

Other population group 8545 2390 8545 239 8545 2390

Percent rural 7926 2732 7899 2762 7912 2748

Living in another country 331 650 331 65 331 650

Socioeconomic status 005 097 -002 102 001 100

Exhibit 22 Homicide rate in Honduras by month

78

We used census data to control for several community-level structural factors Principal

components analysis was used to reduce some of the data into a summary measure Exhibit 23

presents the results of the component loadings One component was extracted that we labeled

socio-economic status (SES) This component exhibited high loadings for percent female

headed household percent unemployment and percent completing primary education

Population was used as our exposure variable and population density (per 1 km) residential

stability percentage of the population that is rural percentage of the population that is

indigenous (verify) and the number of residents immigrated to the United States served as our

control variables Population density and outmigration were logged to address skewness in

these two measures

Exhibit 23 Factor Analysis for Socioeconomic Status

Eigenvalue 236093

Variables Factor Loadings

Female-Headed Households 07977

Unemployed 09269

Primary Education 09303

Findings

The results of our t-test analysis are shown in Exhibit 24 It shows that there was a significant

decline in the homicide rate at the municipal level before and after the gang truce The

homicide rate prior to the truce was 697 per 100000 population and following the truce the

homicide rate was 566 homicides per 100000

79

Exhibit 24 T-test on National Data by Month (n= 27 Months)

Homicide Rate Pre Truce Post Truce Change

Mean 6972346 5663592 -1308754

SE 01677361 01151668 02009224

plt 001

Next we examined the effect of the truce through ARIMA regressions The first model in Exhibit

25 does not include an effect of the temporal trend In this model we again find that homicides

declined significantly in the period following the truce We then employed the ARIMA model

again but this time included a variable (month) to control for the temporal trends in the data

Model 2 in Exhibit 25 shows that when we controlled for temporal trends the impact of the

truce we observed was no longer significant Our findings suggest that homicide did not decline

as a consequence of the gang truce but instead the decline in homicides was part of a longer

term historical trend that was independent of the truce

Exhibit 25 Results from the ARIMA Models (n= 27 Months)

Homicide Rate Model 1 Model 2

Effect SE Sig Effect SE Sig

Truce -1309 0350 -0906 0595

m2 -0623 0481

m3 -0153 1447

m4 -0074 0607

m5 -0104 0450

m6 -0050 0466

m7 -0002 0516

m8 0003 0404

m9 0024 1633

m10 -0396 0522

m11 -0754 2424

m12 0837 0366

Time -0030 0047

Intercept 6973 0254 7258 0472

Autocorrelation Coefficient 0010 0408 0111 0249

plt 001

80

Last we used a fixed effects multi-level negative binomial regression to predict the homicide

rate with population as an exposure variable and controlled for the socio-demographic

characteristics of each community The results are presented in Exhibit 26 The only significant

variable in the analysis was the impact of municipal level population stability Specifically we

found that homicide rates increased in areas where residential mobility was high Once again

our analysis showed that time had a negative effect indicating that homicides were decreasing

in all areas over the study period The gang truce itself was unrelated to the decline in

homicides

Exhibit 26 Results from the Multilevel Negative Binomial Models (n= 27 Months)

Homicide Rate Model 1 Model 2

Effect (SE) Sig Effect (SE) Sig

Fixed Effects

Homicide Rate 0004 0004

(0002)

Truce -0119 -0129

(0073) (0072)

Time -0008 -0007

(0005) (0005)

Month 1 0024 0024

(0059) (0059)

Month 2 -0107 -0107

(0062) (0062)

Month 3 -0041 -0041

(0063) (0063)

Month 4 -0033 -0033

(0065) (0065)

Month 5 -0002 -0002

(0067) (0067)

Month 6 (reference)

Month 7 0027 0027

(0051) (0051)

Month 8 0023 0023

(0057) (0057)

Month 9 0036 0036

81

(0056) (0056)

Month 10 -0051 -0051

(0057) (0057)

Month 11 -0031 -0031

(0058) (0058)

Month 12 0187 0187

(0056) (0056)

Born in the same municipality -0013

(0003)

Pecent rural -0003

(0002)

Socioeconomic status 0029

(0044)

Percent dominant population 0001

(0002)

Density (ln) -0010

(0050)

Living in another country (ln) 0001

(0061)

Intercept -10042 -10051

(0061) (0058)

ln(Alpha) -2368 -2367

(0148) (0148)

Random Effects

Var(Truce Effect) 0066 0066

(0024) (0025)

Var(Intercept) 0388 0309

(0049) (0042)

Cov(truce effect intercept) -0024 -0006

(0027) (0025)

plt 001

82

Conclusions

Violent crime in Honduras is widespread With a homicide rate of about 86 per 100000

population Honduras is one of the most violent nations in the world (United Nations 2013)

This compared to an average homicide rate around the world of 62 per 100000 people and

about 27 per 100000 in Central America (United Nations 2013) Given the high rate of violence

in Honduras it is clear that new and innovative ways of reducing violence should be considered

This case study gave an overview of the implementation and impact of the gang truce

experience in Honduras in 2013

The goal of the truce was to significantly reduce the number of homicides In order to assess

the impact of the gang truce in Honduras a pre-test post-test longitudinal quasi-experimental

design was used Census and homicide data were merged at the municipal level to asses

whether the truce had an impact on homicides controlling for population characteristics and

the natural trend in violent crime

Overall the findings suggest that while the homicide rate in Honduras was on a slight

downward trend the gang truce itself was unrelated to any homicide reductions That is the

gang truce had no measurable impact on homicides in Honduras Given the drastic reductions

achieved in El Salvador and the fact that Hondurasrsquo gang truce was a replication of El

Salvadorrsquos the following discussion will examine two important differences between the two

countries gang truces First the implementation of the truce in Honduras did not appear to

obtain trust between all parties involved and did not achieve any notable short-term

deliverables That is the implementation was not robust Second some suggest that the gangs

in Honduras do not have the organizational sophistication to be able to control their members

on the streets rendering them incapable of carrying out any truce agreements

First the implementation of the truce in Honduras did not appear robust The communication

between the parties was weak and none of the parties completed any significant actions as part

of the truce The Catholic Church and OAS served as the bridge between the two main gangs in

Honduras MS13 and 18th Street and the executive branch of the government The ldquopeace

83

processrdquo as it was called in Honduras was initiated in May 2013 After the general election in

November 2013 the newly appointed administration declared that it would no longer support

the truce process with the gangs As a result the truce was short-lived and the governmental

support for the effort shifted with the change in leadership

Though the negotiations were largely based on the good will of a well-respected negotiator

few tangible incentives were offered during the process The negotiations from the beginning

involved discussions about large scale social programming Given the short time frame these

goals in hindsight were unrealistic Naturally there was little trust between the parties at the

beginning of the negotiations and without any quick tangible deliverables from either side the

truce never really materialized

The second challenge to implementing a successful gang truce in Honduras might have been

related to the nature of the gangs themselves It was unclear whether the gangs possessed the

level of cohesion and hierarchical leadership required to make some measures feasible If the

gangs do not have the organizational capacity to control their members on the streets any

agreements that come from the negotiations would be difficult to implement In the earlier

case study of the Salvadorian experience it was suggested that the successes in that country

were related to one of the gangrsquos organizational capacity to impose control of its members The

gangs in Honduras might be different There is at least some evidence that MS 13 in Honduras

might have less organizational leadership (Bossworth 2013) less control over turf (Latin News

Daily Report 2013) and have been more organizationally fractured (Dudley 2013) than their

counterparts in El Salvador The gangrsquos ability to operate as an efficient organization can greatly

impact the outcome of the truce process In general we know that gangs do not have very high

levels of organizational sophistication (Decker Katz and Webb 2008 Decker Bynum Weisel

1998) It might be that gangs in Honduras are more the norm in terms of organizational

capacity

In summary the 2012 truce negotiations in Honduras did not produce any measurable

reductions in the homicide rate The gangs wanted to speak to Honduran society and they even

84

preferred to speak with the governmental authorities but they never interacted with either

The negotiation process seemed to end as quickly as it started The gangs did not deliver with

lower rates of violence and the government did not provide social programs It would be safe to

say that a robust gang truce did not materialize in Honduras The implementation of the truce

seemed to struggle for two primary reasons First the mediators were not able to accomplish

any quick wins to build trust between the parties involved The commitment level on all sides

was not clear throughout the process Second it is not clear whether the gangs in Honduras

have the organizational capacity to control members on the street as would be required to

carry out an effective gang truce In short the 2013 gang truce in Honduras was unsuccessful

85

Conclusions Policy Implications and Recommendations

The purpose of this report was to systematically examine the effectiveness of gang truces Gang

truces have been widely implemented but rarely evaluated Of those gang truces that have

been evaluated little attention has been given to why and how they came into existence In

this report we reviewed prior research on gang truces and presented case studies of gang

truces implemented in El Salvador Jamaica and Honduras for the purpose of understanding

the negotiation processes undertaken with and between gangs and other stakeholders We

were interested in identifying the actors involved in the negotiations the goal(s) of the

negotiations and the strategies employed to carry them out Most importantly however we

wanted to determine whether the gang truce resulted in a reduction in the number of

homicides Each case study offers lessons learned that are unique to their particular

circumstances and when considered together provide direction to policymakers on the

benefits and risks of implementing gang truces

The case studies presented here constitute the most comprehensive evaluations of gang truces

to date Existing documents were used to collect information about the processes associated

with each gang truce Many of these documents included such items as peer reviewed articles

books and reports The majority of these documents were collected over the Internet

requesting documents from those close to the truce and searching library databases Related

the case studies made use of articles obtained from local newspapers The newspaper articles

were not only intended to provide a historical record of the development of each gang truce

but also to provide additional insight into the various external forces that might have impacted

the gang negotiations Because the newspaper serves as a forum for the community to speak

about its concerns newspaper articles also provided a rich source of data on how those in the

community felt about the gang truce Accordingly the newspaper articles offered a different

view of the problem and offered different opinions as to how a gang truce should or should not

be implemented We also conducted a small number of in-depth qualitative interviews with

key informants These data were collected to supplement existing documents and to clarify

issues associated with the negotiation processes This included but was not limited to

86

questions pertaining to identifying the actors involved in the negotiations the goal(s) of the

negotiations and strategies employed to carry out negotiations The interviews were intended

to obtain information from those who possessed first hand knowledge about the gang truce in

each nation

We examined the impact of each truce using official data We first performed a simple t-test

comparing the homicide rates before and after the truce However as discussed above this

technique has limitations The most severe of which is that even if the test was significant it

would be difficult to determine whether the difference was due to the gang truce or a natural

change over time in the outcome We addressed this limitation by using time series models

whereby the homicide rate for the community was modeled as a function of time with truce

period indicators included to measure the effect of the truce net of the temporal trends These

models were estimated with ARIMA techniques Supplemental models were also employed to

examine and control for factors other than the truce that might have impacted homicide over

the study period

Summary of Findings Related to the Implementation of a Gang Truce in the Three Sites

We found that the implementation of gang truces have a number of common characteristics

The first is that in each case a community was experiencing an uncharacteristically high number

of gang related homicides over a fairly lengthy period of time The continued high level of

violence in each case resulted in the community placing strong pressure on the government in

general and the justice system in particular to respond to the problem quickly and effectively

In each case they had first attempted to control gang violence through suppression oriented

strategies and these strategies were found to be ineffective over the intermediate and long

term In turn each communityrsquos inability to exercise traditional informal and formal social

control to decrease levels of violence became self evident to the public and government This

resulted in both the state and community to seek (or participate in) an alternative strategy in

which negotiators would formally andor informally work with gang leaders to establish a truce

that would reduce gang homicide

87

Key stakeholders involved in the negotiation and establishment of each gang truce were fairly

similar In each of the cases examined the gang leaders of the largest and most violently

involved gangs were willing to consider participating in negotiations that could lead to a truce

In each of these cases it was clear that the gangs not only sought to collaborate with the

negotiators for the purpose of reducing violence but perhaps more importantly were seeking a

means in which to gain greater more positive recognition in the community and to reap some

form of benefit to themselves their members and possibly their community In each case

while not always formally involved government officials were at a minimum made aware of

negotiations and in some cases solicited the assistance of third partyrsquos to broker an agreement

between stakeholders In each case it was at least implicitly understood that the government

would ldquolistenrdquo to the gang leaderrsquos expectations of the government and what theymdashthe gang

leaders--had to offer in exchange We found that when the government was no longer willing

to ldquolistenrdquo to or collaborate with negotiators the truce processes ended abruptly Negotiators

were typically comprised of a very small group (ie 2-3) of individuals who were perceived to

be ldquohonest brokersrdquo In El Salvador and Honduras this included a high ranking Catholic Church

official a leader from an international diplomatic organization (ie OAS) and other neutral

parties In Jamaica this included a quasi-governmental organization that had been established

for the purpose of brokering negotiations between gangs to reduce violence and the local

university which had access to staff who were perceived to be neutral but had an interest in

reducing violence due to its proximity to the university

The strategies used to execute each gang truce were similar but yet importantly different They

were similar in that each involved a team of negotiators working to identify common goals to

be achieved and identifying tangibles that could be delivered to the gang leaders gang

members and their community in exchange for the gang achieving their stated goals They

were different however in terms of the structure of the delivery of each parties promise to the

other In Honduras and Jamaica it appears that gang leaders committed to reducing gang

violence in exchange for general promises made by the negotiators for example that

substantial public works programs would be implemented for the goal of reducing

unemployment among gang members and the community In both of these cases it required

88

the government to develop and deploy large scale social programming in a very quick period of

timemdashsomething that neither government had a strong record of demonstrating In El Salvador

negotiators employed a strategy of the gang leaders promising to deliver immediate changes in

gang member behavior for immediate administratively natured changes by the government

For example in exchange for a reduction in gang violence the government agreed to relocate

imprisoned gang leaders to less restrictive prisons and provide them some privileges Following

the successful execution of the first part of the truce which resulted in near term success for

both parties they began to negotiate issues that would take a longer period of time for the

gangs and government to deliver Our findings suggested that some promised deliverables need

to be easily and quickly delivered early in the process so that trust increases between both

parties Stakeholders only have a brief period of time to provide promised benefits before trust

is lost and that tangible benefits need to be delivered in weeks or months not years

Summary of Findings Related to the Impact of Gang Truces in the Three Sites

Our findings suggest that El Salvadorrsquos gang truce had a significant and dramatic impact on the

homicide rate The mean number of monthly homicides declined from about 354 prior to the

truce to about 218 following the truce for a net decrease of about 136 homicides per month

Our forecast showed that between March 2012 and June 2014 the truce had saved about 5501

lives However there was no significant change between the pre-truce and post-truce periods

in the number of thefts extortions robberies rapes and auto theftsrobberies We also found

that the gang truce did not result in a homogenous decline in violence across municipalities

About 61 percent of municipalities experienced a decline in homicides but the decline in

violence varied substantially between municipalities We examined this issue further by parsing

out the relative influence of the number of MS13 and 18th Street gang members on the street

and in prison from each municipality Our analyses indicated that following the truce the

number of MS13 and 18th Street gang members on the street in a municipality was not

significantly related to a decline in homicide but the number of imprisoned MS13 and 18th

Street gang members was associated with a significant change in homicides following the gang

truce In particular the number of imprisoned MS13 gang members was associated with a

89

significant decline in homicides following the gang truce and the number of imprisoned 18th

Street members was associated with a significant increase in homicides following the truce

In Jamaica our initial findings showed that the gang truce might be an effective mechanism for

reducing violence Bivariate analyses showed a significant decline in homicides immediately

after the truce was implemented This explains the work previously published by policymakers

researchers and news reporters Upon further examination of the data however comparing

change in the target and comparison areas and accounting for temporal trends we found that

the decline in homicide was part of a larger nationwide decline in violence and that the gang

truce was not responsible for the decline The only significant effect that we uncovered was the

possibility that homicides were displaced outside the target area for a brief period of time but

then returned to normal

Our findings from Honduras told a similar story as Jamaica Initial analysis showed that the

number of homicides on average declined across municipalities following the gang truce

Specifically the mean number of homicides declined by 13 per 100000 population with 687

homicides per 100000 population on average occurring in each municipality prior to the truce

and 566 homicides per 100000 population on average occurring in each municipality after

the truce However after we examined the effect of the truce through the ARIMA model and

included a variable (month) to control for the temporal trends in the data the impact of the

truce we observed in our bivariate analysis was no longer significant Our findings as in

Jamaica suggested that the decline in homicides was not the consequence of the gang truce

but instead the decline in homicides was part of a long term decline in homicides due to

exogenous factors

The Potential Benefits and Consequences of a Gang Truce

Over the last several years there have been a number of naturally occurring experiments

involving gang truces in a variety of nations in various regions of the world Findings from

evaluations of gang truces are mixed As noted above in El Salvador the gang truce could be

characterized as highly effective at least for the two years following the truce It is worth

mentioning that even after the truce breakup homicides rates while above truce levels

90

continue slightly below pre-truce levels In Jamaica and Honduras the gang truce had no impact

on violence In Los Angeles and Trinidad there was evidence that violence decreased for at

least ninety days but then increased substantially beyond those rates observed prior to the

gang truce (see the introduction section of this report for this discussion) As a consequence it

appears that the potential for long term consequences might out weigh the potential for short

term benefits Only one study site(El Salvador) demonstrated a truce having a substantial and

long term impact on violence Others conversely demonstrated the truce had no impact or

increased violence over the long term In fact a number of scholars have noted that gang truces

are likely to result in a boomerang effect with gang violence increasing over the long run

because of enhanced cohesion within the gang (Klein 1995) Maguire (2013) notes that when

government officials negotiate a truce with gangs they might ldquoinadvertently be acknowledging

gangs as legitimate social entitiesrdquo (p 11) This in itself might increase cohesion among gangs

which has been found to be associated with increased levels of criminality (Decker et al 2008

Klein 1971 Maguire 2013)

Further research is needed to examine how gang truces might impact group cohesion and if it

does whether this in turn results in greater violence Gang truces convey the well-intentioned

image that violence has been addressed and policymakers are doing something about the

problem but researchers need to better understand the probability of a gang truce reducing

violence increasing violence or having no impact This will better position policymakers to

understand the relative risks associated with these types of interventions

Our findings also suggest that while gang truces could be an effective intervention in areas

where gangs are highly structured and organized such as El Salvador they could be counter-

productive in areas where gangs are not as structured and organized Because the vast majority

of street gangs are not well organized (Klein 1995 Spergel 1995) the utility of a gang truce in

reducing violence might be limited Our findings coupled with prior research suggest that gang

interventions need to be tailored to the nature of the gang and its members or it risks

increasing gang violence

Final Thoughts

91

Our analysis suggests that gang truces should only be used as a means of last resort and then

only under certain conditions Given the risks associated with a gang truce communities with

high levels or at least modest levels of formal social control should rely on other more

promising gang control strategies Only when the state has limited or greatly reduced capacity

for social control should a truce be considered Concomitantly a gang truce should only be

considered when a community is experiencing a substantial amounts of gang violence

Communities that are experiencing minimal to modest amount of gang violence may have more

to lose from the establishment of a gang truce than they have to gain Additionally our

findings suggest that a gang truce might only be feasible when gangs are sufficiently organized

to the extent that they have the capacity to regulate memberrsquos behavior In other words gang

leaders must have the ability to reduce their memberrsquos involvement in violence for a gang truce

to work Our findings more concisely suggest that gang truces should only be considered when

there are a great number of gang homicides the state has limited capacity to address the

problem and gang leaders have enough informal social control over their members that they

themselves can substantially control the levels of violence in their community by regulating

their memberrsquos behavior

Recommendations

Gang truces are conjunctural strategies States who suffer from gang-related violence must

establish permanent public policies for crime control and prevention A government that

considers implementing a gang truce should be aware that it cannot become the center

strategy of its public policy for citizen safety

Gang truces should only be used as a means of last resort and then only under certain

conditions Stakeholders must determine whether a process of dialog or negotiation with gangs

is legal ethical and feasible

Stakeholders must anticipate demands that are likely to arise and their response options Some

demands may be easily met such as improved prison conditions Others are much more

difficult and amorphous such as community development through more integrated violence

92

prevention programs (such as those implemented by SolucionES in El Salvador) local economic

development programs or economic reinsertion of ex-gang members

Stakeholders should incorporate immediately achievable and demonstrable deliverables Long-

term goals and promises are unlikely to create the trust needed to sustain a gang truce

Stakeholders must first determine the position in which they are negotiating the incentives that

are possible to deliver and the boundaries and limits they face Gangs are mostly likely to trust

representatives from NGOs community-based organizations and members of the faith based

community as brokers because they are considered more reliably neutral advocates for peace

They need to understand the capacity of the government to deliver promises in a timely

manner

Governments have to make a choice about the visibility and transparency of its participation

This decision needs to be made in the context of the national and local laws the publicrsquos

expectations of transparency and patterns of practices of the past

Governments must be strategic in their support for truce initiatives Some donor funded

programs run by the government prohibit gang member participation and if the government

does not receive approval from the donor it may risk the donor withdrawing its sponsorship of

the program

Governments must ensure an inter-institutional coordination for the management of truces to

avoid the responsibility to be of a single government institution It is necessary to generate or

collect reliable and pertinent data that can be used to analyze and assess the process

It is necessary to implement an effective monitoring system of the truce process similar to that

used in the full report as it can help parties understand what is working and who is delivering

on their promises More specifically the monitoring and truce management system should be

able to identify truce violations and be prepared to respond through the use of legal and

effective practices if stakeholders do not comply

93

Finally it is necessary to develop evaluations of gang truces and monitoring programs and

support violence prevention activities local economic development activities and pilot

programs to support the reinsertion of ex-gang members into society Clearly national

governments municipal governments NGOs and community-based organizations need

increased capacity and resources to discourage the growth of gangs among at-risk youth It is

therefore increasingly important to create economic opportunities for gang members willing to

leave the gangs and find other legal employment Developing and sustaining those

opportunities in nations with high incidences of poverty will require significant international

funding

94

References

Anderson Elijah 1999 Code of the Street New York Norton

mdash 1998 ldquoThe Social Ecology of Youth Violencerdquo In Crime and Justice vol 24 edited by M

Tonry 65-104

mdash 1990 Street Wise Chicago University of Chicago Press

Arce A (2013 May 28) Honduras gangs declare truce ask talks with govt Associated Press

The Big Story Retrieved from

httpbigstoryaporgarticlehonduras-gangs-declare-truce-seek-talk-govt

Archibold Randal 2012 ldquoGangsrsquo Truce Buys El Salvador a Tenuous Peacerdquo The New York Times

Originally published online on August 27 2012

Axelrod R (1986) ldquoAn Evolutionary Approach to Normsrdquo American political science review

80(04) 1095-1111

Ayala Edgard 2012 Gangs Back Plan for Violence-Free Districts in El Salvador

httpwwwipsnewsnet201212gangs-back-plan-for-violence-free-districts-in-el-

salvador Found on November 4 2014

Bakrania Shivit 2013 Policy Responses to Criminal Violence in Latin America and the

Caribbean GSDRC Helpdesk Research Report 934 Birmingham UK GSDRC University

of Birmingham

Bargent James 2013 ldquoRise in Disappearances Feeds Doubts Over El Salvador Trucerdquo August 1

2013 Found at httpwwwinsightcrimeorgnews-briefsrise-in-disappearances-in-el-

salvador-feeds-doubts-over-truce on September 1 2014

Battin Sara R Karl G Hill Robert D Abbot Richard F Catalano and J David Hawkins 1998

ldquoThe Contribution of Gang Membership to Delinquency Beyond Delinquent Friendsrdquo

Criminology 36 93-115

Block Richard 2000 ldquoGang Activity and Overall Levels Of Crimerdquo Journal of Quantitative

Criminology 16 (3) 369-83

The Gang Truce as a Form of Violence Intervention

Technical Report

SolucionES Project Cooperation Agreement No AID-519-A-12-00003

Coordination Document prepared by Fundacioacuten Nacional para el Desarrollo - FUNDE Roberto Rubio Fabian PhD ndash Executive Director FUNDE Patricia Valdeacutes ndash Project Coordinator SolucionES-FUNDE Juan Meleacutendez ndash Research Technician Project SolucionES-FUNDE

Authors Charles M Katz PhD Arizona State University Luis Enrique Amaya MS

San Salvador El Salvador CA 2015

ACKNOWLEDGEMENTS

This document was made possible through the generous support of the people of the United States of America

through the United States Agency for International Development under cooperation agreement no AID-519-A-12-

00003 The contents of this document are the sole responsibility of the authors and do not necessarily reflect the

views of the United States Government The research was also funded in part by the ASU Foundation through a

generous gift by the Watts Family The opinions expressed herein are those of the authors and are not necessarily

those of the ASU Foundation or the Watts Family We would also like to acknowledge support from Dr Anthony

Harriott who provided data and contextual support from Jamaica John Maisto who assisted with data acquisition

in Honduras Eric Hedberg who provided statistical assistance and Andrew Fox Cher Stuewe-Portnoff Beto Brunn

Giuliano Perseu Juan Melendez Patricia Valdes Tim Nelson and Lidia Nuno for their helpful comments and

suggestions throughout the project

July 2015

Table of Contents

Page Executive Summary i

Section 1 Introduction 1

Section 2 Gang Truce - The Salvadorian Experience 11

Section 3 Gang Truce ndash The Jamaican Experience 44

Section 4 Gang Truce ndash The Honduran Experience 70

Section 5 Conclusions Policy Implications and Recommendations 85

References 94

Appendix A Systematic Review of the Literature 107

Appendix B Truce Related Media Statements in El Salvador 111

Appendix C August Town Five Year Peace Agreement 121

Appendix D Time Line El Salvador Gang Truce 123

Appendix E Social Violence Prevention Programming 124

i

EXECUTIVE SUMMARY

Background

While there is much literature describing the assumptions issues and effectiveness of crime

suppression (eg Decker 2003 Decker and Reed 2002 Katz and Webb 2006 McCorkle and

Miethe 2002) and prevention strategies (Esbensen and Osgood 1997) much less attention has

been paid to gang intervention programming particularly gang truces Little is known about

how often gang truces occur what conditions give rise to them the role of third parties in

brokering them their transformative effects and their effectiveness In this policy brief

sponsored by SolucionES1 and conducted by FUNDE a member of the SolucionES Alliance with

Arizona State University we systematically evaluate gang truces including reviewing prior

research and presenting evidence on the effectiveness of gang truces that have been

implemented in El Salvador Honduras and Jamaica for the purpose of identifying lessons

learned should other governments or donors wish to support gang truces in these or other

countries

Summary of Findings

We found that the truce in El Salvador resulted in a reduction in homicides that was not the

product of other trends or temporal factors Thus the truce in this country had a short-term

effect in reducing the lethality of violent crime though not necessarily in other dimensions of

insecurity By contrast the truces in Jamaica and Honduras resulted in no impact on violence

The Jamaican and Honduran experiences therefore mirror the results of prior gang truces that

1 SolucionES is multifaceted violence prevention program being implemented by an Alliance of five leading Salvadoran non-profit organizations who have come together to prevent crime and violence The Alliance members are Fundacioacuten Nacional para el Desarrollo (FUNDE) Fundacioacuten Salvadorentildea para la Salud y el Desarrollo Humano (FUSAL) Fundacioacuten Crisaacutelida (known locally as Glasswing) Fundacioacuten Salvadorentildea para el Desarrollo Econoacutemico y Social (FUSADES) and Fundacioacuten Empresarial para el Desarrollo Educativo (FEPADE) with partial funding from USAIDEl Salvador Together these five organizations have widely-recognized expertise in education health community development economic development research and youth leadership they are bringing their combined synergy and strengths to prevent crime and violence in El Salvador SolucionES is using a three-pronged strategy to prevent crime and violence in El Salvador 1) Strengthen municipal crime and violence prevention capabilities and actions 2) Increase social investment by the private sector to prevent crime and violence and 3) Research publish and disseminate findings to inform decision-making on crime and violence prevention This is one of 10 planned policy-oriented studies The full study is available on request

ii

have been studied to various extents including those in Los Angeles and Trinidad and Tobago

Important differences in how the various truces were negotiated may explain the different

results and one important difference ndash the ability of government and non-gang community

stakeholders to promise and immediately produce measurable deliverables ndash appear to be

especially significant

Prior Evaluations of Gang Truces

Little research to date has examined the effectiveness of gang truces and much of the research

that does exist has been restricted to the field of public health Studies of a gang truce in South

Central Los Angeles between the Crips and Bloods by Cotton (1992) and Ordog et al (1993

1995) found temporary reductions in the number of homicides and gunshot wounds during the

truce but these studies failed to point out that despite about a 35 percent decrease in

homicides for the first three months of the truce

homicides then doubled in months four through

eleven compared with the pre-truce period

Similar findings were reported in Trinidad and

Tobago (Maguire Katz and Wilson 2013) where

violence declined for a brief period of time

(again for about three months) but then

increased substantially over the long term (12

months) These results suggest that gang truces

may produce short-term benefits yet result in

long-term adverse consequences

Additionally these studies have suffered from a

variety of flaws including 1) poor theoretical

assumptions about gangs including often

ignoring that they might have the organizational

structure and internal discipline to enforce their

truces 2) relatively weak evaluation designs that fail to account for other potential causes for

What is a Gang Truce

A gang truce is a nonviolent resolution to a larger conflict between groups that has an impact on general levels of violence and other forms of criminality within a community (Ordog et al 1993 1995 Whitehill et al 2012) It differs from conflict interruption resolution or mediation efforts which seek to rapidly intervene typically through outreach workers or violence interruption specialists in episodic violent events between groups in a community Gang truces often involve dialogue and negotiations between multiple parties (eg gangs government NGOrsquos religious organizations) that seek to recalibrate the norms of conflict within and between groups for the purpose of reducing or eliminating violence and other crime

iii

or broader trends related to violence reductions 3) a failure to examine the processes involved

in creating gang truces 4) a failure to examine unintended consequences of the truces beyond

their impact on gun violence and 5) a lack of sophisticated statistical analysis capable of

discerning overall trends cyclic patterns outliers and turning points

Overview of Study Design

This policy brief examines gang truces implemented in El Salvador Jamaica and Honduras

including the processes undertaken with and between gangs and other stakeholders We

collected information about the processes associated with each gang truce from a variety of

sources including peer reviewed articles books reports local newspaper articles and

interviews of persons with first-hand knowledge about the gang truce in each nation We

examined the impact of each truce using official data We first performed a simple t-test

comparing the homicide rates before and after the truce We then employed time series

models not used in other studies to evaluate homicide rates as a function of time with truce

period indicators included to measure the effect of the truce net of the temporal trends We

also employed supplemental models to examine and control for factors other than the truce

that might have affected homicides over the study period

Research Findings

Implementing a Gang Truce

The gang truces studied have a number of common characteristics First in each case a

community was experiencing an uncharacteristically high number of gang related homicides

over a fairly lengthy period of time which resulted in each community placing strong pressure

on the government in general and the justice system in particular to respond to the problem

quickly and effectively Second each community had first attempted unsuccessfully to control

gang violence through suppression-oriented strategies Third each communityrsquos inability to

exercise traditional informal and formal social control to decrease levels of violence became

self-evident to the public and government This resulted in all involved stakeholders wanting to

seek (or participate in) an alternative strategy in which brokers would formally andor

informally work directly with gang leaders to establish a truce that would reduce homicides

iv

Key stakeholders involved in the negotiation and establishment of each gang truce were also

somewhat similar In each case examined the leaders of the largest and most violently

involved gangs were willing to participate In each of these cases it was clear that the gangs

sought to collaborate with the brokers not only for the purpose of reducing violence but

perhaps more importantly as a means to gain greater more positive recognition in the

community and to reap some form of benefit to themselves their members and perhaps their

community In each case while not always formally involved government officials were at a

minimum made aware of negotiations and in some cases solicited the assistance of third

parties to broker an agreement between stakeholders In each case it was at least implicitly

understood that the government would ldquolistenrdquo to the gang leaderrsquos expectations and offers

We found that when the government was no longer willing to ldquolistenrdquo to or collaborate with

brokers the truce processes ended abruptly Brokers were typically comprised of a very small

group (ie 2-3) of individuals who were perceived to be ldquohonest brokersrdquo In El Salvador and

Honduras this included a high ranking Catholic Church official a leader from an international

diplomatic organization (ie OAS) and other neutral parties In Jamaica this included a quasi-

governmental organization that had been established for the purpose of brokering negotiations

between gangs for the purpose of reducing violence as well as the local university which had

access to staff who were perceived to be neutral but had an interest in reducing violence near

the university

The strategies used to execute each gang truce were generally similar but had important

differences Each involved a team of brokers working to identify common goals to be achieved

and tangibles that could be delivered to the gang in exchange for the gang achieving the stated

goals They differed by how each partyrsquos promise was delivered In Honduras and Jamaica gang

leaders traded violence reduction pledges for long-term government and social changes such

as the development of substantial public works programs to reduce unemployment In

Honduras and Jamaica the government was asked to develop and deploy large scale social

programming in a short time framemdashsomething for which neither country had a strong track

record In El Salvador by contrast brokers secured promises for immediate changes in gang

behavior in exchange for feasible immediate deliverables from the government For example in

v

exchange for a reduction in gang violence the government agreed to immediately relocate

imprisoned gang leaders to less restrictive prisons and provide them some privileges Following

the successful execution of the first part of the Salvadoran truce which resulted in near term

success for both parties they began to negotiate broader issues that would take longer for both

sides to deliver Our findings suggest that some promised deliverables need to be easily and

quickly delivered by both parties early in the process to achieve trust and serve as a first test

of gang leadersrsquo ability to deliver Stakeholders have only a brief period of time to provide

promised benefits before trust is lost meaning that tangible benefits need to be delivered in

weeks or months not years

The Impact of a Gang Truce

El Salvador We found that El

Salvadorrsquos gang truce had a definite

impact on the homicide rate The

mean number of monthly homicides

declined from about 354 prior to the

truce to about 218 following the

truce for a net decrease of about

136 homicides per month Our data

show that between March 2012 and

June 2014 the truce had saved

about 5501 lives (see exhibit 1)

From a hypothetical stand it is

possible to make the assumption that a number of these deaths averted could have been

transformed in disappearances and therefore they were not counted within the official

homicide statistics However the results from the analysis point out that the number of

disappearances was not significantly related with the change in the global behavior of

homicides Additionally over the period of analysis (January 2010 to June 2014) there was no

significant change between the pre-truce and post-truce periods in the number of thefts

extortions robberies rapes and auto theftsrobberies

Exhibit 1 Forecast of homicides without gang truce

vi

We also found that the gang truce did not result in a homogenous decline in violence across

municipalities About 61 percent of municipalities experienced a decline in homicides but the

decline in violence varied substantially between municipalities We studied this issue further by

examining the impact of the initiative ldquoFree Violence Municipalitiesrdquo and it was found that the

behavior of violence in those municipalities was not significantly linked to the initiative but

rather to the general dynamic of the truce process nationwide Additionally we parsed out the

relative influence of the number of MS13 and 18th Street gang members on the street and in

prison from each municipality Our analyses indicated that following the truce the number of

MS13 and 18th Street gang members on the street in a municipality was not significantly

related to a decline in homicide but the number of imprisoned MS13 and 18th Street gang

members from the municipality was In particular the number of imprisoned MS13 gang

members from a municipality was associated with a significant decline in homicides in that

municipality following the gang truce and the number of imprisoned 18th Street members

from a municipality was associated with a significant increase in homicides in that

municipality following the truce These findings lend support to the idea that MS13 is more

organized than the typical street gang and that imprisoned MS13 gang members exhibit strong

influence over their fellow gang members on the street Our findings also suggest however

that the gang truce had a boomerang effect in municipalities with high numbers of imprisoned

18th Street members implying that 18th Street might not have as much organizational capacity

to regulate violence on the streets as MS13 The truce provided incarcerated MS13 and 18th

Street gang leaders an opportunity to negotiate with high-ranking officials and influential

diplomats including representatives of the Organization of American States This may have

increased their legitimacy inside and outside of their gangs It appears that MS13 was able to

exert its span of control over the communities in which they had influence and they were able

to deliver on the terms of the gang truce negotiations In the case of 18th Street however

incarcerated gang members may not have had the same organizational capacity for

communicating and carrying out directives In fact a review of the gang truce indicated that

there was a conflict taking place between two factions within 18th Street Consequently the

organizational structure and culture of 18th Street might be more diffuse than that of MS13

vii

and its leadership structure might not be as strong because of the internal fractures within the

gang This might further explain why homicides increased in 18th Street communities The

internal fractures within the 18th Street gang may have resulted in intra-gang violent conflict

that was largely contained within 18th Street controlled territories

Jamaica In Jamaica at first glance our impact findings appeared to show that the gang truce

might be an effective mechanism for reducing violence Bivariate analyses showed a significant

decline in homicides immediately after the truce was implemented This explains the work

previously published by policymakers researchers and news reporters Upon closer

examination of the data however comparing change in the target area to other areas in

Jamaica and accounting for temporal trends we found that the decline in homicide was part of

a larger nationwide decline in violence and that the gang truce was not responsible for the

decline The only significant effect that we uncovered was the possibility that homicides were

displaced outside the target area for a brief period of time but then returned to previous

levels

Honduras Our impact findings from our analysis of data from Honduras tell a very similar story

as Jamaica Initial analysis showed that the number of homicides on average declined across

municipalities following the gang truce Specifically the mean number of homicides declined by

12 per 100000 population from an average of 687 per 100000 population in each

municipality before the truce to an average of 566 thereafter However after we examined the

effect of the truce through time series analysis and included a variable (month) to control for

the temporal trends in the data the impact of the truce we observed in our bivariate analysis

was no longer significant Our findings as in Jamaica suggest that the decline in homicides was

less a consequence of the gang truce than of a broader short to medium term trend And

importantly in both Jamaica and Honduras the respective governments failed to deliver on

gangsrsquo demands for large-scale social and employment programs

Conclusions and Policy Implications

Over the last several years there have been a number of naturally occurring experiments

involving gang truces in a variety of nations and regions Findings from evaluations of gang

viii

truces are mixed In El Salvador the gang truce could be characterized as an effective short

term strategy to reducing homicides It is worth mentioning that despite homicides rates are

above truce levels they continue slightly below pre-truce levels In Jamaica and Honduras the

gang truce had no short or medium-term impact on overall violence In Los Angeles and

Trinidad y Tobago there was evidence that violence decreased for at least ninety days but then

increased substantially beyond those rates observed prior to the gang truce Thus the

effectiveness of the truce in El Salvador appears to be isolated and must be evaluated within

the context of other truces that have failed to reduce homicide violence Policy makers must

evaluate whether the conditions that allowed short term effectiveness of the gang truce in El

Salvador (such as the ability to promise and deliver immediate results) exist in other violent

areas before evaluating whether a truce strategy might be appropriate And they should be

heavily cautioned that the potential for long term negative consequences might outweigh the

potential for short term benefits

Indeed it is important to note that a number of scholars have noted that gang truces are likely

to result in a boomerang effect with gang violence increasing over the long run because of

enhanced cohesion within the gang (Klein 1995) Maguire (2013) notes that when government

officials negotiate a truce with gangs they might ldquoinadvertently be acknowledging gangs as

legitimate social entitiesrdquo (p 11) This in itself might increase cohesion among gangs which has

been found to be associated with increased levels of criminality (Decker et al 2008 Klein 1971

Maguire 2013) Hence it is important to consider the fact that gangs are illegal groups in El

Salvador and it should be cautious when carrying out dialog or negotiation processes with

them Further research is needed to examine how gang truces might impact group cohesion

and if it does whether the cohesion created could be effectively directed toward more

productive non-violent endeavors Gang truces convey the well-intentioned image that violence

has been addressed and policymakers are doing something about the problem but unless the

truce is implemented in a manner and under conditions where immediately achievable results

can be promised delivered and measured there remains a significant chance that the truce will

fail or worse yet backfire Thus it is imperative that any type of concession made by

ix

Governments to gangs within a truce framework should be transparent so that all sectors of

society have certainty that every action is being done within the existing rule of law2

In the case of El Salvador the truce arises from to the absence of effective public policies and

practices for violence control and prevention The truce was planned as a strategy to reduce

gang-related homicides During the process different organizations got involved including

religious (facilitating and protecting human rights) non-government (managing and facilitating

dialog and negotiation processes) international (providing funding for insertion programs) and

government (facilitating and providing certain conditions for dialog and negotiation)

organizations Some of the concessions that the Government provided in order to achieve a

reduction of homicides were within the law but others generated confusion and they seemed

to be close to the legal or socially acceptable limits This fact along with the poor transparency

of authorities in the management of the practice with public media and public weakened the

process and postponed its continuity

The present study suggests that gang truces should only be used as a means of last resort

and then only under certain conditions Given the risks associated with a gang truce

communities with high levels or at least modest levels of formal social control should rely on

other more promising gang control strategies such as pulling levers (ie Boston Ceasefire)

community oriented policing and the Gang Resistance Education and Training (aka GREAT)

program Only when the state has limited or greatly reduced capacity for social control should a

truce be considered Concomitantly a gang truce should be considered as an alternative only

when a community is experiencing a substantial amount of gang violence Communities that

are experiencing minimal to modest amount of gang violence may risk more from the

establishment of a gang truce than they have to gain Additionally our findings suggest that a

gang truce might only be feasible when gangs are sufficiently well organized to be able to

regulate their membersrsquo behavior and cause their members to behave less violently In El

22 Pragmatically this issue is complicated On the one hand transparency is a foundational element within a democracy and is necessary to ensure proper oversight of the government On the other hand it might not be possible to implement a gang truce with too many actors having a voice Policymakers might consider creating a policy that allows such negotiations take place but requires particular actors (such as a judicial body) to be informed of the process to ensure transparency and adherence to the rule of law

x

Salvador there is evidence of the strong organizational structure of gang MS13 Among other

factors the magnitude of its membership the chain of command from its leaders in prison and

the discipline of its leaders in the streets seemed confirmed On the contrary the

organizational structure of gang Barrio 18 ndash divided in two factions fighting over the leadership

ndash showed to be a less stable counterpart within the truce

Finally dialog and negotiations processes with or between gangs must have the capacity to

promise and deliver immediate benefits to the gangs that gang members can see or experience

in order to secure their continued participation in the truce as well as the capacity to monitor

and respond to truce violations Most importantly any effort aimed at reducing violence is

important and should be examined and assessed but it must have a transparent foundation

especially when it affects population rights as a whole

General recommendations

Gang truces are conjunctural strategies States who suffer from gang-related violence must

establish permanent public policies for crime control and prevention A government that

considers implementing a gang truce should be aware that it cannot become the center

strategy of its public policy for citizen safety

Gang truces should only be used as a means of last resort and then only under certain

conditions Stakeholders must determine whether a process of dialog or negotiation with gangs

is legal ethical and feasible

Stakeholders must anticipate demands that are likely to arise and their response options Some

demands may be easily met such as improved prison conditions Others are much more

difficult and amorphous such as community development through more integrated violence

prevention programs (such as those implemented by SolucionES in El Salvador) local economic

development programs or economic reinsertion of ex-gang members

Stakeholders should incorporate immediately achievable and demonstrable deliverables Long-

term goals and promises are unlikely to create the trust needed to sustain a gang truce

Stakeholders must first determine the position in which they are negotiating the incentives that

are possible to deliver and the boundaries and limits they face Gangs are mostly likely to trust

xi

representatives from NGOs community-based organizations and members of the faith based

community as brokers because they are considered more reliably neutral advocates for peace

They need to understand the capacity of the government to deliver promises in a timely

manner

Governments have to make a choice about the visibility and transparency of its participation

This decision needs to be made in the context of the national and local laws the publicrsquos

expectations of transparency and patterns of practices of the past

Governments must be strategic in their support for truce initiatives Some donor funded

programs run by the government prohibit gang member participation and if the government

does not receive approval from the donor it may risk the donor withdrawing its sponsorship of

the program

Governments must ensure an inter-institutional coordination for the management of truces to

avoid the responsibility to be of a single government institution It is necessary to generate or

collect reliable and pertinent data that can be used to analyze and assess the process

It is necessary to implement an effective monitoring system of the truce process similar to that

used in the full report as it can help parties understand what is working and who is delivering

on their promises More specifically the monitoring and truce management system should be

able to identify truce violations and be prepared to respond through the use of legal and

effective practices if stakeholders do not comply

Finally it is necessary to develop evaluations of gang truces and monitoring programs and

support violence prevention activities local economic development activities and pilot

programs to support the reinsertion of ex-gang members into society Clearly national

governments municipal governments NGOs and community-based organizations need

increased capacity and resources to discourage the growth of gangs among at-risk youth It is

therefore increasingly important to create economic opportunities for gang members willing to

leave the gangs and find other legal employment Developing and sustaining those

opportunities in nations with high incidences of poverty will require significant international

funding

xii

References

Cotton Paul 1992 ldquoViolence Decreases with Gang Trucerdquo JAMA 268(4) 443-44

Decker Scott H 2003 Policing Gangs and Youth Violence Belmont CA Wadsworth

Decker Scott H and W Reed 2002 Responding to Gangs Evaluation and Research

Washington DC National Institute of Justice

Decker Scott H Charles M Katz and Vincent J Webb 2008 ldquoUnderstanding the Black Box of

Gang Organization Implications for Involvement in Violent Crime Drug Sales and

Violent Victimizationrdquo Crime and Delinquency 54 153-72

Esbensen Finn-Aage and D W Osgood 1997 ldquoNational Evaluation of GREATrdquo US

Department of Justice Office of Justice Programs National Institute of Justice

Katz Charles M and Webb Vincent J 2006 Policing Gangs in America New York Cambridge

University Press

Klein Malcolm W 1995 The American Street Gang New York Oxford University Press

mdash 1971 Street Gangs and Street Workers Englewood Cliffs NJ Prentice-Hall

Maguire Edward Charles Katz and David Wilson 2013 ldquoThe Effects of a Gang Truce on Gang

Violencerdquo Unpublished paper Washington DC American University

Maguire Edward 2013 Research Theory and Speculation on Gang Truces Woodrow Wilson

International Center for Scholars in Washington DC

McCorkle R and T Miethe 2002 Panic The Social Construction of the Street Gang Problem

Upper Saddle River NJ Prentice-Hall

Ordog Gary J W Shoemaker Jonathan Wasserberger and Michael Bishop 1995 ldquoGunshot

Wounds Seen at a County Hospital Before and After a Riot and Gang Truce Part Twordquo

Journal of Trauma-Injury Infection amp Critical Care 38(3) 417-19

Ordog Gary J Jonathan Wasserberger Julius Ibanez Michael Bishop Eduardo Velayos

Subramaniam Balasubramanium and William Shoemaker 1993 ldquoIncidence of Gunshot

Wounds at a County Hospital Following the Los Angeles Riot and a Gang Trucerdquo Journal

of Trauma 34 779-82

Whitehill J M Webster D W amp Vernick J S 2012 ldquoStreet conflict mediation to prevent

youth violence Conflict characteristics and outcomesrdquo Injury prevention injuryprev-

2012

1

Introduction

Given the increasingly devastating individual family and community effects of gang activity

over the past several decades an increasing body of literature has focused on gangs gang

members and gang activity A core theme running throughout this body of literature is that

gang members are significantly more likely to be the offenders and victims of violent crime than

non-gang members (Battin Hill Abbott Catalano and Hawkins 1998 Esbensen Winfree He

and Taylor 2001 Katz et al 2011 Huff 1998 Curry et al 2002 Miller and Brunson 2000

Miller and Decker 2001 Pyrooz et al 2012 Curry et al 2001) and disproportionately affect

neighborhood levels of crime and violence (Block 2000) These findings have been robust in

that they have been repeatedly found regardless of research methodology (ie use of official

data self-report data observational data) (eg Curry 2000 Decker 1996 Deschenes and

Esbensen 1999 Esbensen et al 2001 Gordon et al 2004 Katz Webb Schaefer 2000 Webb

Katz Decker 2006) or research setting (ie North America Europe Asia South America

Australia) (Klein and Maxson 2006) As a consequence it should not be surprising that

policymakers academics and community leaders have focused much of their attention on

developing responses to address community gang problems

Suppression strategies have been the favored public policy response to gangs since the 1980s

(Venkatesh 1999 Spergel et al 1995) Suppression strategies typically rely on focusing

criminal justice resources on gang members through such practices as targeted and enhanced

police patrols intelligence databases vertical prosecution and enhanced sentences for those

convicted (Katz and Webb 2006) Suppression strategies are based on deterrence theory and

are founded on the principal that swift certain and severe penalties for those involved in gangs

and gang activities will necessarily result in fewer individuals joining gangs and will deter people

from engaging in gang crime (Klein 1995) By the early-to-mid 1990s as gang problems

continued to proliferate and policymakers sought alternative gang control strategies gang

prevention programming flourished (Papachristos 2013) Gang prevention programs were

aimed at the general youth population or focused on at-risk youth or neighborhoods Gang

prevention programming was based on the premise that by reducing risk factors and increasing

2

protective factors prevention specialists could inoculate youth from gang membership

(Esbensen 2000) Prevention programming has most frequently come in the form of school-

based prevention programming such as the Gang Resistance Education and Training (GREAT)

program where students are exposed to a curriculum designed to reduce both participation in

gangs and gang crime as well as the SolucionES3 program which engages youth and their

families through integrated community-based and school based programs

While a burgeoning body of literature has developed describing the assumptions issues and

effectiveness of suppression (eg Decker 2003 Decker and Reed 2002 Katz and Webb 2006

McCorkle and Miethe 2002) and prevention strategies (Esbensen and Osgood 1997) much less

attention has been given to examining gang intervention programming This might largely be

the consequence of the absence of such programming over the past 30 to 40 years By the

1980s policymakers no longer believed that social intervention approaches were an effective

strategy to control gangs and gang problems Although gang intervention strategies took many

forms they were based on the assumption that gang membership is the by-product of a socially

deprived community and that the values and norms of gang youth can be influenced and

directed toward those of mainstream society As such intervention programs often focused on

diverting youth from gangs or sought to minimize the consequences of gangs and gang activity

Gang intervention strategies include crisis intervention dispute resolution street-level

counseling and youth outreach (Spergel 1995) However many policymakers community

organizations and academics have argued that such approaches not only did not reduce gang

activity but may have lead to increased group cohesiveness which in turn may have lead to

increased delinquency (Klein 1971 Spergel 1995) More recent research has yielded similar

results For example a number of studies examining replications of Chicago CeaseFireCure

3 SolucionES is a Global Development Alliance in which $20 million is provided by USAID and SolucionES Alliance

members will contribute an additional $22 million dollars during a 66-month period starting July 2012 SolucionES is being implemented by an Alliance of five leading Salvadoran non-profit organizations who have come together to prevent crime and violence The Alliance members are Fundacioacuten Nacional para el Desarrollo (FUNDE) Fundacioacuten Salvadorentildea para la Salud y el Desarrollo Humano (FUSAL) Fundacioacuten Crisaacutelida (known locally as Glasswing) Fundacioacuten Salvadorentildea para el Desarrollo Econoacutemico y Social (FUSADES) and Fundacioacuten Empresarial para el Desarrollo Educativo (FEPADE) Together these five organizations have notable expertise in education health community development economic development research and youth leadership they are bringing their combined synergy and strengths to prevent crime and violence in El Salvador More information on social violence prevention programming is presented in Annex E including a summary of the violence prevention initiatives being implemented by SolucionES

3

Violence which relies heavily on crisis intervention dispute resolution street-level counseling

and youth outreach have found these strategies to either be ineffective or worse increase

levels of violence (Fox et al 2014)

One type of gang intervention gang truces however has received little attention in the

literature This is somewhat surprising given its use and public claims of effectiveness The

purpose of this report is to systematically examine the effectiveness of gang truces Gang truces

have largely been a ldquoblack boxrdquo that is little is known about the frequency of their use

conditions that give rise to them the role of third parties in brokering gang truces the

transformative effects of truces and the effectiveness of gang truces In this report we review

prior research on gang truces and present evidence on the effectiveness of gang truces that

have been implemented in El Salvador Jamaica and Honduras

Theoretical and Policy Rationales for Gang Truces

Much of the concern about gangs over the past twenty years has been their close association

with violence Qualitative and quantitative research has repeatedly found that gangs and gang

members are involved in high levels of serious violent offending Decker (1996) attributes the

gang-violence relationship in part to the collective and normative structure of gangs which

supports and encourages the use of violence both preemptively and in retaliation He further

explains that gang membership encourages the use of violence in retaliation against threats

and attacks whether actual or perceived which results in patterns of inter-gang conflict

characterized by escalating violence Related Cooney (1998) points out that gangs are not all

that different from ldquowarrior societiesrdquo He argues that while gangs are violent the violence

that they engage in is different than that engaged in by non-gang members in that it commonly

takes on a feud-like dimension A perceived slight violation of turf or other disrespectful

action might invoke a shooting which in turn evokes a retaliatory shooting which in turn

results in another retaliatory shooting creating a self-perpetuating cycle of violence

Early research and theoretical work examining gangs and gang conflict suggested that much

violent gang behavior was the function of status management (Thrasher 1927 Whyte 1943

Short and Strodtbeck 1965) Borrowing from the sociology and social psychology literature on

4

impression management theory gang scholars hypothesized that youth place significant

emphasis on image managementmdashseeking to impress their peers and limiting the potential to

embarrass themselves in front of others (Hughes 2005) These early gang theorists postulated

that violence is an instrument used by gangs and gang members to achieve manage and

protect status They contend that gang members often make decisions to become involved in a

violent conflict based on rational processes that weigh the immediate loss or gain of status

within the gang against the relatively small probability of being formally sanctioned by officials

within the criminal justice system (Hughes 2005) A number of researchers have examined the

relationship between status considerations and gang violence and have found the association

to be particularly robust regardless of gender (Campbell 1991 Giordano 1978) ethnicity

(Anderson 1990 1998 Vigil 1988) and location (Jankowski 1991 Spergel 1995) Hughes

(2005) notes that the centrality of status consciousness by gang youth may account for the

existence of the facilitative gang effect that gang scholars have repeatedly observed (ie Gangs

increase the amount of delinquency individuals are involved in beyond that of a group of

delinquent peers (Gordon et al 2004 Thornberry et al 2003)

Another micro-social factor associated with gang violence is group cohesiveness While our

understanding of the relationship between gang cohesiveness and violence is underdeveloped

some scholars have reported a strong relationship between the two In particular Klein and his

associates (1971 1967) and Lucore (1975) have reported that increases in gang cohesiveness

lengthens periods of gang membership and increases members participation in gang crime and

violence Cooney points out that there are strong relational ties between gang members that

necessarily result in increased cohesiveness among members For example gangs are

comprised of neighborhood youth who share common cultural and economic experiences and

have often known one another for lengthy periods of time He also points out that gangs are

groups that have strong self-proclaimed and formalized identities (eg colors symbols names

monikers) and have at least some organizational structure Their sense of group is maintained

by their common understanding of their members and friends and their attachment to their

territory (or neighborhood) Decker (1996) notes that the relatively high level of group

5

cohesiveness exhibited by gangs facilitates both collective behavior and liability (For exception

see Hughes 2013)

Accordingly both micro-social factors (ie status management gang cohesiveness) serve to

augment levels of gang violence and make it difficult for third parties to intervene Violence

within the context of gangs serves as a form of informal social control Gangs and gang

members cannot seek assistance from legitimate institutions of social control to solve conflicts

because they would risk losing status (Anderson 1999) Likewise the collective nature of the

gang not only increases potential offenders and targets of violence but also facilitates at the

group level the need for retributive justice on the behalf of injured members Moreover gang

members desire to impress others with their commitment to the group and use violence to

demonstrate their commitment to their group and to increase their status within the group All

of this results in an increased cycle of gang conflict and violence4

Over the past several decades residents and policymakers have expected the police to address

violent gang conflicts However there are several limitations to the police response to gangs

First as noted above many of the same factors that are associated with violent gang conflict

also limit the effectiveness of the police to have an impact on violent gang conflict Gang

members are not going to contact the police to resolve a conflict because it could result in loss

of status and expose them to the police discovering illegal activities that they are involved in

(Katz 2001 Katz and Webb 2006) Related citizens in neighborhoods with gang problems are

also reluctant to call the police out of fear of gang reprisals (Katz 1997 Webb and Katz 2003)

or because they have a poor perception of the police themselves (Katz Choate and Webb

2002) Third the police response to gangs in most communities is a reactive response to a

specific incident after it has occurred rather than a proactive response to intervene in ongoing

disputes between gangs (Katz and Webb 2003 2006) Most police agencies simply do not have

4 There is anecdotal evidence for these hypotheses Hughes and Short (2005) through field observations and interviews with members of 20 Chicago gangs examined the micro-social context of violent-related incidents Specifically they were interested in the specific social conditions that precipitated violence Their analysis revealed that gang disputes associated with retaliation escalated into violence significantly more often (ie 55 of the time) than disputes associated with normative order violations or an identity attacks which only resulted in violence roughly 33 percent of time

6

the intelligence networks required to intervene in gang conflicts until after they have risen to

relatively high levels (Katz 2003) Fourth police suppression strategies have been linked to

increasing gang cohesiveness and possible increases in gang crime (Klein 1995)

Some policymakers and community activists have proffered that an alternative to reliance on

formal mechanisms of social control such as the police is the gang truce (Spergel 1995) The

goal of a gang truce is to reduce or even eliminate violent conflict between those gangs that are

warring with one another As such compared with other strategies that often seek to reduce

general levels of crime a gang truce has the very specific goal of reducing violence between

two or more gangs that are in conflict with one another Unfortunately the literature provides

little guidance on the theoretical assumptions of why a gang truce should reduce inter-gang

violence Henderson and Leng (1999) hypothesize that at the root of gang truces is the notion

that they involve the renegotiation of existing norms within and between gangs The authors

claim that as a violent dispute escalates between gangs leaders and members are placed in the

situation of appearing weak to both members of their own gang and to members of the rival

gang if they do not respond with the appropriate amount of force or if they were to suggest a

peaceful resolution to the dispute Accordingly Henderson and Leng argue that as the cycle of

violence escalates between two or more gangs behavioral norms shift toward the increased

valuation of violence to resolve the conflict because it is the only option readily available to

them A gang truce on the other hand which is often mediated by a third-party is believed to

break the cycle of violence by providing the gangs involved in the dispute with a cooling-off

period (Spergel 1995) In the interim period new norms of expected behavior within and

between gangs are established In other words the cooling-off period is believed to recalibrate

norms of behavior that are more consistent with the security interests of the gang and its

members (Henderson and Leng 1999)

For many of the reasons stated above a number of communities have participated in gang

truces Gang truces have been observed in the United States Central America and the

Caribbean and gang leaders in other nations are considering its implantation (Fahah 2012)

Unfortunately little is known about gang truces We do not know when they come into

consideration how they are implemented whether they decrease increase or even have a

7

significant impact on violence In the below section we systematically review the existing body

of literature on gang truces Appendix A describes the methodology we employed to conduct

the systematic review

Prior Research on Gang Truces

Little research to date has examined the effectiveness of gang truces and of the research that

has been conducted has been restricted to the field of public health The most celebrated

truce to date was Philadelphiarsquos House of UMOJA 1974 gang truce In preparation of the truce

gang leaders in prison were called upon for their support along with key community leaders

from churches businesses schools and the police When UMOJA called for the gang summit it

was reported that 500 gang members and 75 of Philadelphia gangs attended By the

conclusion of the gang summit a 60-day truce was announced which reportedly resulted in no

gang members being killed over the 60-day period However no systematic evaluation of the

gang truce was conducted (Woodson 1981)

The first attempt to evaluate a gang truce to our knowledge was conducted by Cotton (1992)

who examined the results of a gang truce in South Central Los Angeles between the Crips and

Bloods Data provided by the police department indicated that over the 6-week period when

the truce took place drive-by shootings decreased by 48 compared to the same 6-week

period the prior year decreasing from 162 to 85 Likewise gang-related homicides dropped by

62 from 26 to 105

Ordog et al (1993 1995) examined the effects of a gang truce in Los Angeles using emergency

room admissions data Specifically the authors examined changes in the daily and monthly

number of gun shoot wound (GSW) emergency room admissions before during and after a

gang truce The catchment area for the emergency room was 100 square miles The authors

noted that while they were able to clearly identify the date the gang truce began because of

the media coverage that it received there was no specified date that the truce ended and as a

5 The authors did not report whether the drop in gang-related homicides was citywide or in the South Central Los Angeles neighborhood where the gang truce took place

8

consequence after 12 weeks the authors regarded it as called off for evaluation purposes

because it was no longer being discussed in the media and gun shot wounds began to increase

Student t-tests were used to examine changes before during and after the gang truce Ordog

et al (1995) reported that there were approximately 7 GSW admissions per day the 12-months

preceding the truce compared to 45 GSW admissions per day during the gang truce and 126

GSW admissions per day in the 11-months following the gang truce The authors concluded

that their analysis ldquoclearly showed that the institution of a gang truce had reduced the number

of GSW victims seen in an hellip inner city Level I trauma centerrdquo (p 419)6

However it is important to point out that while the gang truce in Los Angeles did decrease

homicides by about 35 percent for the first three months it then doubled in months four

through eleven compared with the pre-truce period - a pattern that the authors did not

discuss Similar findings were reported in Trinidad and Tobago (Maguire Katz and Wilson

2013) where it was determined that homicides declined for a brief period of time (again for

about three months) but then increased substantially over the long term (12 months) These

results suggest that gang truces may produce short-term benefits yet result in long-term

adverse consequences

While the research examining gang truces shows their potential promise and their potential for

greater harm we believe that the findings should be viewed with caution for two reasons

First some of the implicit theoretical assumptions that gang truces are built on may not be

accurate Prior research on gangs suggests that they have limited organizational structure

(Decker Katz and Webb 2008 Decker Bynum Weisel 1998) and have few formal

mechanisms to influence member behavior For example most gangs do not have formal

leaders do not require members to pay dues and members do not make contributions to the

gang for the purpose of developing the gangs infrastructure (ie guns housing etc) A number

of academics have pointed out that ldquosophisticated gang organizations is still largely a product of

the self- or organizational-interested musings of gang leaders certain police officials academic

6 Similar results were presented by Ordog et al 1993 using the same data but analyzing it over a shorter period of time and aggregating the data by month instead of by day

9

researchers media reporters based on very limited hard datardquo (Spergel 1995 79-80)

Therefore even if a truce was successfully negotiated between members of gangs in conflict

much of the academic research suggests that gang leaders do not have enough control over

members to enforce a truce Related prior research on gangs and gang members suggests that

they are primarily comprised of young people with few ldquodiplomaticrdquo skills (Henderson and Leng

1999) The very same organizational and normative features of gangs that result in gang

violence (ie often bastions of young minority male street youth) are those same features that

most likely limit the likelihood of a gang having the capacity to abide by a gang truce In sum

implicit in the theoretical assumptions of a gang truce is that gangs have the organizational and

cultural capacity to create and maintain (at least for a short while) a truce which is inconsistent

with much prior research on the organizational characteristics of gangs

The second reason we believe that prior evaluations of gang truces should be viewed with

skepticism is that they have relied on relatively weak evaluation designs For example prior

studies examining gang truces have not incorporated the use of comparison areas or control

groups The causes of reductions in gang violence found in previous evaluations might be

many For example the Los Angeles riots took place just before the gang truce that Ordog et al

evaluated Zinzun (1997) reported that gang culture and violence changed briefly but abruptly

following the riots because gangs and gang members in part redirected their anger and focus

toward the police As such the decline in GSW admissions may have been the consequence of

an overall city-wide decline in gang violence in the wake of the LA riots Related prior research

has relied on fairly broadly defined outcome measures such as GSW admissions and general

levels of drive-by shootings Such outcome measures lack specificity in terms of attributing

gang violence to the specific gangs involved in the gang truce Evaluation designs used in

previous studies have also lacked specificity in terms of the catchment area where violent

activity was labeled as gang related For example Ordog et alrsquos (1993 1995) outcome measure

included all emergency room admissions for GSW However the emergency room received

patients from a 100 mile square area surrounding the hospital an area that was most likely

much larger than the gangsrsquo territories involved in the truce

10

Additionally prior research examining gang truces has not examined the processes involved in

the creation of the gang truce No context has been provided in terms of the factors that lead

to the gang truces whether the gangs were pushed or pulled into truces whether outside

parties helped mediate the truces or whether on-going mediation was required to maintain the

truce In other words we still do not know the processes related to the formation of a

successful gang truce Instead prior research and evaluation has treated the gang truce much

like a black box where it is described in very general terms but its details are not revealed

Likewise prior evaluations have not examined the effects of a gang truce beyond its impact on

gun violence A number of prior researchers have argued that mediating such activities serves

to legitimize gang leaders increases gang identity for members and results in greater group

cohesion (Haskell and Yablonsky 1982 Klein 1995) As such we do not know if there are any

unintended consequences resulting from gang truces due to the methodologies chosen for

prior research and evaluation Finally prior research on gang truces has not relied on more

sophisticated statistical analysis that has the capacity to not only discern overall trends but also

cyclic patterns outliers and turning points

THE PRESENT STUDY

The present study seeks to understand the effectiveness of the gang truce when negotiations

occur with or between gangs and with government institutions or civil society organizations

We have briefly reviewed the theoretical and policy rationales for gang truces and the

literature and prior research on this form of gang intervention Next we present three case

studies one from El Salvador another from Jamaica and another from Honduras Each offers

evidence about their impact on violence and lessons learned that are unique to their particular

circumstances In conclusion the case studies are considered together and we present their

collective policy implications

11

Case Study A Gang Trucemdash The Salvadorian Experience

Over the past two years the Salvadoran gang truce brokered by an array of local policymakers

and international development organizations has attracted national and international

attention Violence in El Salvador has been at an all time high with a rate of 692 homicides per

100000 population in 2011 (UNODC 2014) making El Salvador one of the most violent nations

in the world (Parkinson 2014) Since 1992 government and non-government actors have been

responding to the rising tide of gang violence using traditional suppression strategies such as

increased policing legislative changes and more severe prison sentences These traditional

mechanisms of formal social control however were proving ineffective if not counter-

productive (Perez 2003)

In response to the inadequacy of traditional strategies stakeholders altered their course in an

effort to radically reduce gang violence in the nation Members of the Funes administration led

a group of negotiators comprised of the Catholic Church a former congressman and the

Organization of the American States (OAS) to help frame the conditions for a possible truce

between the MS13 and 18th Street gangs (Umana de Leon and Tager 2014) In March 2012 a

truce was reached The goal of the gang truce was to reduce violence specifically gang-involved

homicides Included in the terms in exchange for the gangs acting to reduce homicides certain

incarcerated gang members were to be transferred to lower security prisons to receive

sentence reductions and special visitation privileges and to be permitted more communication

with those outside of the prison for the purpose of conducting crisis interventions to mitigate

the violence (Salanegra 2012)7 The gang leaders also agreed to no longer recruit children into

their gangs reduce violence against women give up a small number of guns and continue to

participate in negotiations (Seelke 2014 11-12)

7 It is important to note that there were other discussions that took place during the mediations that included as a local publication noted ldquoProhibition Act Gang send the army to barracks end the police operations in controlled areas by gangs repeal of the figure of the witness criteriado (with criteria of opportunity or ldquowitness under a plea agreementrdquo) and a number of improvements in the quality of life of the inmates and its familiesrdquo Source httpwwwsalanegraelfaronetes201209cronicas9612

12

The present case study seeks to understand the role of negotiations with or between gangs and

government institutions or civil society organizations and their impact on violence in El

Salvador The specific objectives of this case study were to (1) Identify and document the

negotiation processes with or between gangs This includes but will not be limited to

identifying the actors involved in the negotiations the goal(s) of the negotiations and

strategies employed to carry out negotiations (2) Identify the impact of negotiations with

andor between gangs on violence and other forms of criminality and (3) Present conclusions

about the strengths and weaknesses of negotiations and their potential for positively or

negatively reducing crime in Salvadoran society

The Salvadorian Gang Truce

While the origins of the Salvadorian gang truce are somewhat unclear some accounts suggest

that in the fall 2011 the Minister of Security (then David Munguia Payes) mentioned the idea of

starting a dialogue between the gangs to Raul Mijango (a former congressman) Shortly

thereafter Mijango spoke to Monseigneur Fabio Colindres the military chaplin about the idea

They started to communicate with gang leaders in January 2012 From the beginning the

Minister of Security stated that he would not personally communicate with gang leaders and

noted that the official position of the government was not to negotiate with offenders But he

did allude to the fact that the gang truce was part of a new strategy to address the nationrsquos

gang problem (Archibold 2012) The Salvadorian gang truce was multi-dimensional involving a

varying number of actors communication styles and tactics In the section below we discuss

the parties involved in the truce the negotiation process and the strategies used by the gangs

and government in furtherance of the truce

Parties involved in the truce There were a series of persons groups and other entities who

directly or indirectly participated in the negotiation process whether they were negotiators or

collaborators From the beginning negotiations between the MS13 and 18th Street gangs were

considered confidential and were held in strict confidence between the gangs and the

negotiators Beginning in early 2012 the first two mediators to serve were a former

congressman (Rauacutel Mijango) and a priest (Fabio Colindres) These individuals were able to gain

13

the trust of the gangs and also had open communications with the government and

community

As the process moved forward however church leadership changed its initial position and

withdrew its support of the truce This left the priest to participate as a private individual no

longer serving as a representative of the church The mediators had some early success

however as the mediators changed the lines of communication became less clear and less

reliable Still they played an important role in the process Finding individuals whom the gangs

would trust and who could speak for government and community institutions was challenging

Further as government ministers and presidents changed it became less clear whether or not

the government was participating in the dialogue between various parties much of the

confusion in negotiations was related to understanding the governmentrsquos official role in the

process

In the beginning the negotiating parties included the two mediators and MS13 and 18th Street

the two predominant gangs in El Salvador MS13 is the largest gang in the country with about

250 cliques throughout the nation For the most part its first- and second-generation leaders

were in prison There are municipalities in which only MS13 operates and therefore its

members never clashed with 18th Street adversaries MS13 members are largely involved in

extortion violence and intimidation of the public The MS13 leadership has been shown to

have greater control of its organization than most its members have been more disciplined

and despite its size the gang seems to have sufficient and effective communication

mechanisms between cliques and its leadership (ldquoranflardquo or ldquoruedardquo)

The 18 Street gang is somewhat smaller than MS13 Long before the truce process began

factions existed within 18th street which had become deeply involved in internal conflicts over

leadership structure Those differences had become more pronounced over time leading to the

establishment of two movements known as the ldquoSurentildeosrdquo (Southerners) and the

ldquoRevolucionariosrdquo (Revolutionaries) that participated in the peace process but under different

14

representation Despite their differences however for the negotiation process the two

fractions of 18th street established similar demands between themselves and with MS138

As noted above the governmentrsquos involvement in the truce process was never clearly

articulated and it varied with ministerial changes The Funes Administration specifically the

Ministry of Justice and Public Security (MJPS) had been engaged in the negotiation process

early on In 2013 however with the change in MJPS leadership this also changed Tensions

grew over the role of the government in the truce negotiations According to the mediators the

new Minister of Security wanted to dismantle the process and for this reason he blocked the

mediatorsrsquo entry to the prisons where they had been meeting with gang leaders The change in

government leadership brought changes in strategy As with all negotiations mutual trust was

the cornerstone When the personnel and their agendas changed so did the trust in the peace

process

Late in 2012 as negotiations began to focus on the relationship between the gangs and the

community other organizations were called upon to help facilitate this interaction A technical

committee was created to oversee progress towards the violence reduction goal this was

announced at the end of August 2012 The committee moved forward with support from the

Organization of American States (OAS) It coordinated actions intended to improve the living

conditions of incarcerated persons to reintegrate into society those who had served their time

in prison and to prevent violence and provide assistance to victims of violence The committee

operated as a mechanism for coordinating the negotiation process with technical support from

the United Nations Development Program (UNDP)‒El Salvador Interviews with experts

indicated that Fundacioacuten Humanitaria with support from several organizations like Interpeace

and others had also supported the operations of the mediators and served as an important

point of reference for the gangs

In each territory where agreements were made government representatives from some

municipalities were invited to participate in the process as key implementation stakeholders

Between December 2012 and January 2013 the municipalities of Ilopango Santa Tecla

8 httpsalanegraelfaronetes201110cronicas5645

15

Sonsonate and Quezaltepeque became among the first to become part of what later became

known as the Sacred Municipalities (aka violence free municipalities) By November 2013

several other municipalities had become part of this group La Libertad Apopa San Vicente

Puerto El Triunfo Nueva Concepcioacuten Ciudad Delgado and Zacatecoluca

The gangs and the government offered several incentives to one another The gangs offered a

number of actions to reduce the incidence of crime (ie cease of hostilities zero homicides

stop extortions robberies and recruitment of children and youths into gangs peace at the

schools) and the government offered to guide public investments in social policies prioritizing

participating municipalities

The truce process was the product of a convergence of vested interests from different sectors

Several earlier attempts at negotiations had failed likely because of a lack of perceived

authenticity on the part of the actors All parties involved in this truce process came to the

table with well-defined group interests and concerns All sides were experiencing fatigue from

the long history of violence Years of fighting deaths and violence had led the mostly

incarcerated first generation those who were still living and who continued to serve as gang

leaders to seek new ways of exercising social power and influence in prison and the

community

The negotiation process In El Salvador the negotiation process between gangs formally began

on March 8 2012 by a decision of the Government to favor the transfer of gang leaders

imprisoned out of the countryrsquos only maximum security prison located in the municipality of

Zacatecoluca That decision was made by the Minister of Justice and Public Security as part of

an agreement between a team of mediators who served as advisors to the Ministry and gang

leaders primarily from MS13 and 18th Street The negotiations were referred to as a ldquotruce

between gangsrdquo and were characterized by the mediators as ldquoa peace processrdquo or a ldquocease of

hostilities between gangsrdquo

Publicly the gang truce was known as an agreement between gangs and not between gangs

and the government because of fear of how those outside the negotiations might interpret the

governmentrsquos actions From the beginning therefore the government never fully

16

acknowledged its participation in the peace process (through representatives) with the gangs

and that reticence was a major impediment throughout the process This was largely because

of the difficulty in communication between the multiple parties which was done through

mediators and necessarily resulted in additional logistical complexity throughout the

negotiations Regardless it is important to recognize that the government played a decisive

and central role in the origin facilitation and promotion of the peace process

In terms of processes interviews with gang leaders indicated that they would communicate

their ldquodemandsrdquo to the mediators who in turn would communicate the gang leaderrsquos demands

to the government The same process was used for the government to communicate with the

gang leaders After an agreement was reached the parties would make proclamations that

emphasized the agreed upon terms of the negotiations These often took the form of public

statements made through local newspapers For example there were more than twenty public

statements made by the gangs where they clearly articulated that their intention was to

reduce the harm that their acts of violence were causing themselves and the communities (For

a detailed discussion of the public statements see Appendix B) In these public statements the

gangs acknowledged the need for their groups to change Both the mediators and the gangs

made clear that the gangs expected the government to respect and respond to their demands

given the good-faith actions that the gangs had already taken

The government on the other hand did not make any formal public statements about their

end of the agreements reached through the negotiations From the outside it appeared that the

government did not develop a formal or consistent policy regarding the truce and instead

assumed that they could achieve a reduction in homicides through the negotiation of increased

prison privileges for gang leaders At the close of the Funes administration the peace process

and negotiations initiated within its framework had reached a plateau There they remained

waiting for the new government to step in and take action as President-elect Salvador Saacutenchez

had announced his support for the truce during the electoral process Sanchez however

withdrew that support when he assumed the presidency in June 2014 Indeed under the

recently elected Sanchez administration the government changed course and became less

17

willing to engage the gangs believing that the government should not negotiate with criminal

organizations

Indeed it should be noted that responses resulting in increased gang influence have been a

concern over the past several decades--from governmental neglect of the problem in the

1990s through the period of the mano dura and super mano dura of the early 1990s to

2000 and to the present gang truce Many have asserted this claim as truth (ie some

responses will increase the influence of gangs) especially in public discourse This study

however does not directly address this point which may or may not be valid9

Strategies employed during negotiations between the gangs and the government Over the

course of the negotiations between the gangs and the gangs and the government at least two

strategies were employed At the beginning of the negotiations strategies were implemented

similar to that of the historic Salvadorian peace accords Each of the parties with the support of

mediators reached agreements and achieved a resolution to their conflict This model

although useful overlooked an important difference between the current violence in El

Salvador and the civil war of the 1980s In a conflict of civilians with the State the legitimacy of

the ldquoadversariesrdquo arises by the need of the state to recover the rule of law It is certainly

possible to reach peace agreements that resolve conflict between gangs but in the current

case with regard to the government there was no legal or policy justification for executing the

truce process On the contrarymdashgovernment negotiations with a criminal group are relatively

rare in modern democratic societies The lack of a legal or policy framework to work from

limited the governmentrsquos ability to be transparent in its response to gangs and may have

undermined its legitimacy with the public

The second strategy employed by the parties was based on reciprocation and cooperation

between actors Early on it was determined by both sides that if one of the parties abandoned

the peace process or did not hold up its end of an agreement the other party would no longer

participate in the negotiation process (Axelrod 1986) In retrospect during the course of the

peace process the gangs implemented agreed-upon terms and positioned themselves as valid

9 We would like to thank one anonymous reviewer for bringing this issue to our attention

18

partners with the government able to negotiate for what is needed by the government in

exchange for what the gang needed10 For example in public statements the gangs insisted

that they had made a goodwill gesture when they declared a unilateral truce and stopped their

involvement in violence The government responded by transferring incarcerated gang leaders

from a maximum-security prison to ordinary criminal prisons11 Next the gangs offered to

disarm the government responded with increased flexibility on control measures at those

prisons This process continued until the demands from both sides grew in a direction that

challenged each sidersquos capacity and willingness to follow through

For instance mediators through the government asked the gangs to end the extortion of

businesses and individuals which are the primary means of subsistence for Salvadorian gangs

The gangs asked that the government reciprocate by eliminating the ldquowitness under a plea

agreementrdquo from the criminal procedural law which is one of the main weapon laws used for

sentencing gang members for complex crimes such as extortion Each of these requests were

more than the parties could carry out Exhibit 1 outlines the different ldquooffersrdquo and ldquodemandsrdquo

made by each party The offers and demands are divided into those that were believed to be

relatively simple and were ldquoofferedrdquo by the party on its own volition and those that were

believed to be more complex and were demanded by the other party

Exhibit 1 Simple demands and concessions made by gangs and the government

Gang concessions Government concessions

Reduce the homicides by more than 50 Transfer the leaders to prisons with lesser security

Hand over 500 firearms to the authorities Allow visits of the children of the gang members

Reduce violent actions at schools Allow night intimate visits

Stop killings at the prisons Allow entry of fast foods

Do not murder custodians police soldiers or their family members

Facilitate the entry into the prison of gang members let out of prison

Give opportunities to some gang members to withdraw

Suspend the searches by the armed forces on persons at the prisons

10 Source httpwwwsalanegraelfaronetes201209cronicas9612 11 This was useful to facilitate the coordination and the communication between the gang members in prisons and in the streets

19

Make it more flexible for new social plans by government to be established in the communities controlled by gangs

Improve the conditions and health assistance at the prisons (eg place tvrsquos to improve mental health of inmates)

Do not burn buses do not kill bus drivers or collecting agents in buses

Allow the Red Cross and journalists to enter into the prisons

The second agenda presented in Exhibit 2 shows the more complex demands that were

not offered by one side or the other but rather were demands placed on the other party These

demands were such that they required a higher level of authority to negotiate in order to

implement the demanded action These demands were considered critical for the peace

process to continue and the delayed responses on both sides stalled the progress of the truce

and led to the parties questioning the legitimacy of the other side all these demands have yet

to be attained The mediators recognized at the end of 2012 that some of the slow progress

was related to the difficulty of making the transition from offers that could be executed

relatively simply to the more complex demands made by each party One of the gang leaders

summarized the issue ldquoWe are not looking for television sets while all of our people continue

living like shit we are not going to try to do everything that is in our power to decrease as much

violence as possible for one television set There are things we are very clear about this [points

to a television set in the room] is a right that the law grantsrdquo

Exhibit 2 Complex demands and concessions made by gangs and the government

Demands by Gangs Demands by Government

Eliminate the figure of witness under a plea agreement from the Criminal Law

Stop homicides and extortions indefinitely

Create international commission to investigate the human rights violation cases of the gang members and their family members on the part of the State (PNC and FA)

Share the information regarding the whereabouts of the brunt of extortion money whether in country or abroad

Institutionalize external and professional surveillance regarding the behavior of the PNC as regards investigations and gang member arrests

Progressive dismantling nation-wide of the clique structures and turfs

Army should stay in its barracks and definitely suspend their participation in public security tasks

Permanently suspend the orders to murder State security and justice agents and their family members (police soldiers judges prosecutors)

20

Clearing of judges prosecutors and police involved in corruption cases against youths in conflict with the law

Permanently suspend the murders extortions and harm caused to public transportation resources and their workers

Maintained a sustained improvement on the conditions of the prisons

Permanently suspend the recruitment of children and youths and hostilities to educational centers

Guarantee working opportunities for the gang members and their family members through specific programs at the municipalities

Share the information about providers of drugs and illegal arms

Television sets were one of the concessions made by the government from the simple list The

gang leader recognized that their intentions in the negotiations extended beyond improving

prison conditions however the government did not have support nor established mechanisms

for carrying out tasks derived from the more demanding list Further as the government

transitioned to new presidential leadership many of the more simple concessions had already

been made and the new government would have to address more complex demands if

negotiations were to continue

The Salvadorian gang truce is remarkable for several reasons First a number of policymakers

and researchers have claimed that the truce saved a large number of lives and was perhaps the

most successful gang truce in the Western Hemisphere Second the Salvadorian gang truce is

somewhat unique in that it might have lasted substantially longer than any other successfully

negotiated gang truce Understanding the temporal impact of the truce is important to

understanding its future potential Third the Salvadorian gang truce is important because a

number of other counties have sought to replicate it For example following the perceived

(and perhaps real) success of the Salvadorian gang truce the nations of Honduras Belize and

Guatemala instituted or tried to institute a similar type of truce Understanding the impact of

the Salvadorian gang truce will further help us understand the potential impact of such a

process on violence in these other nations

Methods

We examined the impact of the gang truce by merging four separate data sets First data from

the 2007 El Salvador Population and Housing Census provided municipal level measures of

21

percent unemployed percent male aged 10-29 percent female headed households ethnic

composition in-migration income percent urban percent households rented and percent

who had a high school education These data were obtained directly from El Salvadors Ministry

of Economy Second municipal level crime data (ie homicide extortion rape theft robbery

and auto theftrobbery) were provided by the National Civil Police (PNC) by month and year for

the period between January 1 2010 and June 30 2014 Disappearance data was also provided

by the police aggregated by year and municipality Third police gang intelligence data was

provided by the National Civil Police (PNC) Specifically we received the number of police

recorded MS 13 and 18th Street gang members by municipality in 2011 the latest year for

which these data were made available Last we acquired 2011 prison gang intelligence data on

the number of imprisoned MS 13 and 18th Street members by municipality from the

Salvadorian National Bureau of Prisons All four data sources were linked together using a

unique municipality identification number and month Collectively they provide an

opportunity to examine the impact of the gang truce in El Salvador and understand several

competing explanations for any changes that might have occurred

The geographic unit of analysis for the present study is the municipality According to the 2007

El Salvador Population and Housing Census there were 14 departments divided into 262

municipalities (the equivalent of a county in the USA) in El Salvador Of the 262 municipalities

ten (38) of the municipalities were eliminated from the analysis because of missing data

Measures

Dependent variables The dependent variable in the present study is the monthly number of

homicides in each El Salvadoran municipality expressed as the number of homicides in

municipality i at time t Homicide data were collected from January 1 2010 through June

30 2014 for a total of 54 months This includes data for a period of 26 months prior to the gang

truce and 28 months following the implementation of the gang truce As shown in Exhibit 3

there were a total of 14148 homicides over the study period with each municipality averaging

371 homicides per month

yit

22

Exhibit 3 Descriptive statistics

n mean sd

Urban Opportunity Factor 252 111 95

male 10-29 years old 252 1960 171

female-headed household 252 3433 499

unemployed 252 1177 686

Racialethnic heterogeneity 252 17 13

of prisoners MS13 252 1579 3960

of prisoners 18th Street 252 1268 4096

of prisoners MS13_spatial weight 252 1579 3960

of prisoners 18th Street _spatial weight 252 1268 4096

MS13 on the street 252 4146 10646

18th Street on the street 252 2513 8622

Violence free municipality 252 04 20

Truce 14148 371 783

Monthly homicide rate per 1000000 14148 51 49

Independent variables Several independent variables were included in our model Our

measure for the intervention is a dummy variable in which the value of 0 is used to represent

the pre-intervention months and 1 represents the post-intervention months We also

included a dummy variable for each community that participated in the violence-free

municipality initiative (sanctuary municipalities) to address issues of external validity (eg

multiple treatment inference) In other words municipalities that participated in the

violence-free initiative might have experienced or participated in something that either

enhanced or detracted from the impact of the larger gang truce The initiative called by some

the second phase of the gang truce took place about 8 to 9 months after the original gang

truce was negotiated in March 2012 In participating sanctuary municipalities gang members

agreed to stop violence and crime in exchange for a reduction in police operations and night

raids12 The municipalities that participated in the initiative included Santa Tecla

Quezaltepeque La Libertad Ilopango Ciudad Delgado Apopa Sonsonate San Vicente

Zacateatecoluca Puerto El Triunfo and Nueva Concepcioacuten

12 See httpwwwcispesorgblogviolence-free-cities-inaugurated-as-second-phase-of-gang-truce

Tt

Si

23

We included several independent variables related to the presence of gangs by gang for each

municipality Specifically we included count variables on the number of MS13 and 18th Street

gang members who were on the streets and in prison by municipality As noted before the

gang truce was made between a relatively small number of imprisoned gang leaders from both

MS13 and both fractions of 18th Street who agreed to stop street-level gang violence on the

condition that they would be transferred to lower security prisons and granted special

privileges

We wanted to understand the impact of the truce in the context of gang presence

Municipalities with high numbers of MS13 and 18th Street members whether they were in

prison or on the streets should have experienced a greater reduction in homicides because of a

greater span of control over these communities Accordingly municipalities with low numbers

of MS13 and 18th Street members should have experienced a lesser reduction in homicides

because of a limited span of control over these communities As Exhibit 3 (above) shows on

average municipalities had about sixteen MS13 and thirteen 18th Street members in prison

and forty-one MS13 and twenty-five 18th Street members on the street We examined whether

gang members in El Salvador were randomly distributed and found that there was spatial

clustering in the number of MS13 and 18th Street members who were in prison In order to

control for the clustering we included the spatial lag (weight) of the MS13 and 18th Street

members in prison

Control variables A series of control variables known through prior research to be related to

violence in communities were added using the census data We began with municipal-level

measures of percent unemployed percent male aged 10-29 percent female-headed

households ethnic composition in-migration income percent urban percent households

rented and percent with a high school education Initial analysis found that five community

variables were highly correlated and loaded on the same factor As seen in Exhibit 4 these

included the percentage of persons who had moved there from another municipality (in-

migration) average income in the municipality percent of the population living in an urban

area percent of houses that are rented and percent of residents with at least a high school

education We labeled this factor Urban Opportunity

24

Exhibit 4 Factor loadings for urban opportunity

Variables Factor

loadings

In Migration Percent of population moved in from another municipality 0638

Income Average monthly income per household (colones) 0886

Urban Percent of population that is urban 0845

Rented Percent of households rented 0761

Education Percent of residents who have at least a high school education 0742

Percent of variance 6071

Eigenvalue 341

Extraction method Principal Axis Factoring

We also calculated a measure of ethnic-heterogeneity from relevant census data Ethnic

heterogeneity which varies from 0 to 1 was calculated by taking one minus the squared

proportions of the population in each ethnic group (White Mestizo other) As with the

percentages of male population that is 10-29 years old female-headed households and

unemployed persons ethnic heterogeneity has consistently been associated with violence in

general (Kubrin and Weitzer 2003) and with gang violence in particular (Katz Maguire and

Roncek 2002 Rosenfeld Bray and Egley 1999)

Analytic Strategy

National-level analysis Our analysis began with a simple t-test of the mean number of

homicides at the national level before and after the truce providing the most basic omnibus

test of an effect We then presented our time series model to estimate the effect net of

seasonality and temporal trends This analysis included a set of simple ARMA (autoregressive

and moving average) models with two lag periods and one period of a moving average Initial

analysis of the number of homicides by month indicated that partial temporal autocorrelations

existed for two lags The first model used only data prior to the truce This model included a

linear time trend variable and dichotomous variables for each month (except January) We used

this model to forecast the expected number of homicides for the truce period This series of

analyses was for illustrative purposes only as the number of data points used in the forecast

25

was too small to provide meaningful confidence intervals The second ARMA model employed

all data from January 2010 through June 2014 and included a dichotomous indicator for the

truce period The effect of this variable was the average change in the number of homicides

net of seasonality (months) and temporal trends

Municipality multilevel models The analysis at the municipal level presented four challenges to

a typical regression model First since our outcome has a highly skewed distribution with

varying levels of exposure (population) we employed a generalized linear model to capture the

correct distribution In this case we employed a negative binomial distribution rather than a

Poisson model due to the over-dispersion in our outcome created by analyzing monthly

municipal datamdashthat is there were several months and municipalities where no or very few

homicides occurred

The second challenge was our need to measure the between-municipality variation of pre-truce

homicide rates and the program effect A fixed-effects model would have been inappropriate

because it would not have allowed us to estimate these variance components Thus we

employed a multilevel random effects generalized linear model

The third challenge was temporal autocorrelation Analysis at the national level indicated

partial autocorrelations in the first and second lags Although a generalized model does not

allow for auto-correlated residuals we addressed this by entering in the model two lags of the

homicide rate

The fourth and final challenge to this estimation was spatial autocorrelation An examination of

the Moranrsquos I and autocorrelation coefficients (exhibit not shown) indicated a low level of

autocorrelation for each month but many months were still statistically significant Thus we

estimated spatial lags of the homicide rate and entered them into the model

Accordingly we analyzed the data using random effects negative binomial models regressing

the rate of homicide on various months on predictors with both temporal and spatial lags For

clarity we present the model using HLM (Bryk et al 1996) notation At the first level (time) we

estimated the log number of homicides using a negative binomial distribution with predictors

26

that included the truce period calendar month linear month trend a

one period lag of the homicide rate a two-period lag of the homicide rate a

spatial lag of the homicide rate and an over-dispersion parameter which has a

Gamma distribution of Thus the final level 1 model is

At the municipality level (level 2) the intercept is a function of the presence of the

Violence Free Municipality program the log number of MS prisoners the log

number of 18th Street prisoners the log of the spatial lag number of MS prisoners

the log of the spatial lag of the number of 18th Street prisoners the log of the

number of MS members the log of the number of 18th Street prisoners the

control variables detailed above and the random effect Thus the level 2 model for the

intercept is

Also at the municipality level (level 2) the effect of truce is a function of SAFE the

log number of MS prisoners the log number of 18th Street prisoners the log of

the spatial lag number of MS prisoners the log of the spatial lag of the number of 18th

Street prisoners the log of the number of MS members the log of the number

of 18th Street prisoners and the random effect Thus the level 2 model for the

truce effect is

TRUCEt mt TIMEt

HRt-1i HRt-2i

HRti vti

Gamma1

aa

aelig

egraveccedilouml

oslashdivide

mti = expb0 i + b1iTRUCEt + lmmt

m=1

11

aring + l12TIMEt

+l13HRt-1i + l14HRt-2i + l15HRti + vti

eacute

euml

ecircecircecirc

ugrave

ucirc

uacuteuacuteuacute

exp vti[ ] ~ Gamma1

aa

aelig

egraveccedilouml

oslashdivide

b0 i

SAFEi PMSi

P18thi

PMSi P18thi

MMSi M18thi

Ci u0 i

b0 i = g 00 +g 01SAFEi +g 02 ln PMSi +1( ) +g 03 ln P18thi +1( ) +g 04 ln PMSi +1( )

+g 05 ln P18thi +1( ) +g 06 ln MMSi +1( ) +g 07 ln M18thi +1( ) + pC

C

aring Ci + u0i

b1i SAFEi

PMSi P18thi

PMSi

P18thi MMSi

M18thi u1i

27

The key parameters of our analysis are in this expression The truce effect is dependent on

the log number of MS prisoners among other characteristics Thus the percent

reduction in homicides can be calculated from this model using the following formula

which we plot below (see Quantitative Findings below)

To answer questions about the variation in the effects of the gang negotiations we predicted

best linear unbiased predictions (BLUP) of the Truce random effect which estimates the

between-community variance of the immediate effect of the truce Since we assumed that the

interventions were not evenly spread throughout communities in this study the mixed model is

appropriate to address this issue We conducted this analysis for homicide using a Stata

generalized linear mixed model using full maximum likelihood and an identity covariance matrix

of random effects

Findings

Exhibits 5 and 6 present our findings related to the number of police recorded homicides in El

Salvador from January 2010 through June 2014 The trends in the data and the results of the t-

test show that in the 26 months prior to the gang truce there were on average about 354

homicides per month compared with about 218 homicides per month in the 28 months

following the gang truce Exhibit 7 shows that the gang truce resulted in 5501 fewer homicides

than otherwise would have occurred

b1i = g 10 +g 11SAFEi +g 12 ln PMSi +1( ) +g 13 ln P18thi +1( ) +g 14 ln PMSi +1( )

+g 15 ln P18thi +1( ) +g 16 ln MMSi +1( ) +g 17 ln M18thi +1( ) + u1i

b1i

ln PMSi +1( )

1- exp g 10 + ln PMSi +1( )g 12( )eacute

eumlugraveucirc( )acute100

u1i

28

Exhibit 6 Number of homicides in El Salvador by month with ARIMA Model Fit

Exhibit 5 Change in homicides from pre-truce to post-truce

Obs Mean SD 95 confidence interval

Pre-truce period 26 35442 4200 33745 37138

Post-truce period 28 21835 6361 19369 24302

Change -13606 10638 16574

=p lt 05

29

Exhibit 7 Forecast of homicides without gang truce

Next as seen in Exhibit 8 we examined the spatial distribution of the change in the homicide

rate after the implementation of the gang truce The analysis showed that of the 252 analyzed

municipalities 243 (93) experienced a decrease in homicides however within these

municipalities there were wide variations in the degree of the decline For example of the 243

municipalities that experienced a decrease the decrease in the homicide rate varied from

about 59 percent of these municipalities experiencing a 1-74 percent decrease to about 9

percent of municipalities experiencing a 75 percent or higher decrease Additionally a modest

number (n=19 7) of municipalities experienced an increase in their homicide rate

30

Exhibit 8 Percent reduction in homicide rate by number of municipalities

Exhibit 9 presents six negative binomial models for the monthly number of homicides The first

model examines the impact of the gang truce and implementation of the Violence Free

Municipality program and our interaction variable that measures the additive effect of both

the gang truce and the Violence Free Municipalities program on the number of homicides We

found that although the gang truce was associated with a significant decline in homicides the

Violence Free Municipality program was related to a significant increase in homicides13

However countrywide the additive effect of implementing the gang truce and the Violence

Free Municipalities program was associated with a significant decline in homicides

13 To be clear our bivariate analyses showed that the violence free municipalities program was related to a significant decline in homicides For example on the one hand those municipalities that did not participate in the violence free municipalities program experienced a decline in their homicide rate from 406 homicides per 100000 population in the pre-truce period to 321 homicides per 100000 in the post-truce period On the other hand those municipalities that participated in the violence free municipalities program experienced a substantially greater decline in their homicide rate from 752 homicides per 100000 population in the pre-truce period to 392 homicides per 100000 in the post-truce period However after other variables are controlled for in our negative binomial models we found that the violence free municipalities program did not have a positive impact on homicides over and above the gang truce itself

31

Exhibit 9 Negative Binomial models for monthly number of homicides

Model 1 Coefficient (se)

Model 2 Coefficient (se)

Model 3 Coefficient (se)

Model 4 Coefficient (se)

Model 5 Coefficient (se)

Model 6 Coefficient (se)

Homicide - 1 month lag 0010 0010 0010 0010 0010 0010

(0002) (0002) (0002) (0002) (0002) (0002)

Homicide - 2 month lag 0012 0013 0013 0013 0013 0013

(0002) (0002) (0002) (0002) (0002) (0002)

Homicide rate spatial lag 0003 0003 0003 0003 0003 0003

(0001) (0001) (0001) (0001) (0001) (0001)

Gang truce implemented -0483 -0541 -0553 0057 -0544 0564

(0066) (0065) (0064) (0278) (0064) (0279)

Violence free municipality 0497 0183 -0003 -0076 -0037 -0094

(0181) (0167) (0152) (0143) (0153) (0145)

Ln( +1) of MS13 prisoners 0066 0176 0062 0171

(0023) (0028) (0043) (0028)

Ln( +1) of 18th St prisoners 0067 0044 0064 -0043

(0021) (0024) (0021) (0024)

MS13 prisoner spatial lag -0004 0054 -0003 0049

(0045) (0055) (0046) (0056)

18th St prisoner spatial lag 0022 0014 0019 0012

(0027) (0032) (0027) (0032)

Gang truce violence free municipality

-0346 -0294 -0309 -0161 -0304 -0059

(0158) (0156) (0158) (0146) (0158) (0145)

Gang truce Ln( +1) of MS13 prisoners

-0185 -0185

32

(0030) (0030)

Gang truce Ln( +1) of 18th St prisoners

0160 0060

(0025) (0025)

Gang truce Ln( +1) of MS13 prisoners spatial lag

-0139 -0138

(0064) (0064)

Gang truce Ln( +1) of 18th St prisoners spatial lag

0023 0023

(0039) (0039)

Ln( +1) of MS13 on street 0023 0018 0021 0016

(0013) (0012) (0013) (0012)

Ln( +1) of 18th St on street 0043 0039 0040 0037

(0016) (0015) (0016) (0015)

Time 0010 0011 0011 0011 0011 0011

(0002) (0002) (0002) (0002) (0002) (0002)

1month 0064 0065 0067 0067 0067 0067

(0055) (0055) (0055) (0055) (0055) (0055)

2month -0020 -0020 -0022 -0023 -0022 -0023

(0056) (0056) (0056) (0056) (0056) (0056)

3month 0088 0088 0188 0088 0088 0088

(0052) (0052) (0052) (0052) (0052) (0052)

4month -0011 -0011 -0011 -0011 -0011 -0011

(0052) (0052) (0052) (0052) (0052) (0052)

5month 0020 0020 0020 0021 0020 0021

(0052) (0052) (0052) (0052) (0052) (0052)

6month 0000 0000 0000 0000 0000 0000

7month 0036 0036 0036 0036 0036 0036

(0052) (0052) (0052) (0052) (0052) (0052)

33

8month 0047 0047 0047 0046 0047 0046

(0051) (0051) (0051) (0051) (0051) (0051)

9month -0055 -0055 -0057 -0058 -0057 -0058

(0053) (0053) (0053) (0053) (0053) (0053)

10month 0040 0040 0041 0041 0041 0041

(0052) (0052) (0052) (0052) (0052) (0052)

11month 0022 0022 0022 0022 0022 0022

(0052) (0052) (0052) (0052) (0052) (0052)

12month -0040 -0040 -0042 -0043 -0042 -0043

(0053) (0053) (0053) (0053) (0053) (0053)

Urban opportunity factor 0209 0042 0033

(0040) (0040) (0039)

male 10-29 0039 0007 0009

(0021) (0019) (0019)

female-headed household -0006 0002 0000

(0007) (0007) (0006)

unemployed 0001 -0002 -0000

(0006) (0005) (0005)

Racialethnic heterogeniety 0058 -0039 -0087

(0263) (0235) (0230)

Intercept -10571

-10544

-11054 -11618 -11015 -11558

(0064) (0060) (0194) (0241) (0212) (0257)

Ln(Alpha) -2287 -2289 -2293 -2294 -2293 -2294

(0113) (0113) (0114) (0114) (0114) (0114)

Truce random effect 0202 0193 0202 0144 0201 0143

(0039) (0038) (0040) (0031) (0040) (0031)

Intercept random effect 0315 0226 0192 0164 0183 0158

(0047) (0035) (0032) (0027) (0031) (0027)

Cov(Truce and Intercept) -0162 -0099 -0122 -0086 -0114 -0090

(0037) (0031) (0031) (0025) (0031) (0026)

34

The second model includes those variables from Model 1 but also includes variables that

controlled for community-level structural factors In this model the gang truce remains

significantly associated with a decline in homicides but the Violence Free Municipality program

and our interaction variable were no longer related to a reduction in homicide Although the

percentage of residents in a municipality who are male aged 10 to 29 female-headed

households percent unemployed and ethnic heterogeneity were unrelated to changes in

homicide the urban opportunity factor was significantly associated with homicide

Models 3 and 4 included our measures of intervention as well as our measures of the number

of MS13 and 18th Street on-the-street and incarcerated gang members at the municipality level

The analysis showed that while the number of MS13 on the street was unrelated to homicides

the number of 18th Street gang members was associated with an increase in homicide These

analyses also indicated that the number of incarcerated MS13 and 18th Street gang members

from a municipality was associated with a significant increase in homicides We further

examined whether this relationship was associated with the gang truce by including two

interaction variables as presented in Model 4 One measured the interaction between the

number of incarcerated MS13 gang members and the gang truce and a second measured the

interaction between the number of incarcerated 18th Street gang members and the gang truce

The results showed that following the gang truce the number of incarcerated MS13 gang

members was related to a significant decrease in homicides and the number of incarcerated

18th Street gang members was related to a significant increase in homicides

Models 5 and 6 in Exhibit 9 present the same two models (3 and 4) as above but include the

community-level structural variables Model 5 once again shows that the number of MS13

members on the street remains unrelated to the change in homicides and the 18th Street

members on the street was associated with a significant increase in homicide However Model

6 shows that following the gang truce the number of incarcerated MS13 gang members

remained significantly associated with a decline in homicides while the number of incarcerated

18th Street gang members remained significantly associated with an increase in homicides

35

Community-level structural factors were found once again to be unrelated to changes in

homicide rates

Exhibit 10 below further illustrates the relationship between reductions in homicides by

municipality and the municipal-level presence of MS13 and 18th Street gang members in

prison The figure shows that in municipalities with no incarcerated MS13 and 18th Street gang

members no change in homicides occurred following the gang truce However when a

municipality had ten MS13 gang members imprisoned on average that municipality

experienced a 55 percent reduction in homicides When a municipality had fifty MS13 gang

members imprisoned those municipalities on average experienced a 71 percent reduction in

homicide following the truce Conversely our analysis showed that the number of 18th Street

gang members in prison (from a municipality) had a significant and positive impact (that is the

number increased) on homicides following the truce For example if a municipality had ten

18th Street gang members imprisoned on average that municipality experienced a 31 percent

increase in homicides14

14 We examined whether there was an interaction effect between gang members on the street following the gangs truce Analysis for both MS13 and 18th street and their interaction with the gang truce showed no effect

36

Exhibit 10 Impact of the gang truce by number of gang members in prison at the municipal level

The Gang Truce and Crime Displacement

We examined two types of displacement crime type displacement and method displacement

Some policymakers have suggested that although homicides may have declined as a

consequence of the gang truce other forms of criminality or categories of reported crime may

have subsequently increased Crime type displacement occurs when offenders who focus on

one type of crime switch to another form of crime in order to avoid detection or to benefit in

some other way (Eck 1993) In El Salvadorrsquos case some critics of the gang truce have suggested

that as homicides decreased other forms of criminality such as extortion might have

increased substantially due to gang membersrsquo increased freedom to conduct activities inside

and outside of prison (Dudley 2013 Parkinson 2014) Method displacement occurs when

offenders change their tactics or methods of conducting crimes as a consequence of an

intervention (Eck 1993) Following the gang truce in El Salvador some analysts argued that gang

37

members might have begun to hide the bodies of homicide victims to avoid detection and to

ldquoprotect the integrity of the countryrsquos gang trucerdquo (Bargent 2013 1)

As noted above we rely on police data that measured monthly numbers of thefts extortions

robberies rapes and auto theftsrobberies by municipality We also received data on the

annual number of disappearances in each municipality by year Exhibit 11 presents the

descriptive statistics for these variables It shows that the rates of theft and robbery did not

change significantly between the pre- and post-true periods and the rate of extortions

significantly declined The t-tests showed that the rates of rape disappearance and

theftrobbery fromof an auto increased significantly

Exhibit 11 Descriptive statistics for measures of displacement

Mean Std Err [95 Conf Interval]

Theft rate

Pre-truce 1238 049 1142 1335

Post-truce 1213 049 1115 1310

Extortions rate

Pre-truce 353 023 308 398

Post-truce 271 021 229 314

Robbery rate

Pre-truce 544 025 495 593

Post-truce 553 025 504 602

Rape rate

Pre-truce 049 003 043 056

Post-truce 064 005 054 073

TheftRobbery Vehicle rate

Pre-truce 132 022 090 175

Post-truce 165 023 120 210

Disappearance rate

Pre-truce 805 068 670 940

Post-truce 1007 064 882 1132

p lt 05

We further examined the data (excepting disappearance data) similarly to the analyses above

in which we used random effects negative binomial models and regressed the number of

38

crimes (ie theft extortion robbery rape and auto theftrobbery) on various months on

predictors with both temporal and spatial lags Our independent and control variables

remained the same as those used in Model 6 (displayed above in Exhibit 9) We found that

over the study period there was no significant change between the pre-truce and post-truce

periods in the number of thefts extortions robberies rapes and auto theftsrobberies (tables

not shown)

As noted above we also received data on the number of disappearances by municipality and

year but because the data were provided by year there were not enough data points to

examine them temporally Therefore we added this covariate (number of disappearances by

municipality and year) to Model 6 in Exhibit 9 The results of the analysis indicated that the

number of disappearances was not significantly related to change in homicides the gang truce

remained associated with a significant reduction in homicides and our interaction variables

(number of incarcerated MS13 amp 18th Street members truce) remained significantly

associated with homicides

Conclusions

The present study sought to examine the impact of the gang truce on violence and other forms

of criminality We examined homicide data by municipality prior to and after the gang truce

Our outcome variables were obtained from the PNC along with several independent and

control variables obtained from the Ministry of Economy and the National Bureau of Prisons

We analyzed this data using a number of time series and random effects negative binomial

models where we regressed the rate of homicide on various months on predictors with both

temporal and spatial lags and controlled for other contextual factors This type of analysis

allowed us to correct for the number of homicides in one month being related to the number of

homicides in the previous month (ie temporal autocorrelation) Additionally it allowed for

the fact that some communities participated in supplemental interventions (ie violence free

municipalities initiative) which might have impacted the outcome in the same way in a given

month These techniques allowed us to isolate the effects of the gang truce as best possible

We also used data from the PNC that measured by municipality the number of thefts

39

extortions robberies rapes auto theftsrobberies and disappearances to examine the

possibility of the trucersquos impact on crime displacement and diffusion of benefits`

Our findings suggest that El Salvadorrsquos gang truce had a significant and dramatic impact on the

homicide rate The mean number of monthly homicides declined from about 354 prior to the

truce to about 218 following the truce for a net decrease of about 136 homicides per month

Our forecast showed that between March 2012 and June 2014 the truce had saved about 5501

lives As will be discussed further below the analysis suggests that key stakeholders have the

capacity to renegotiate existing norms of violence and that at least some gangs have the

capacity to exert substantial informal social control over their members that can result in

reduced violence

The gang truce also lasted substantially longer than previously evaluated truces Ordog et al

(1995) for example reported that the much publicized gang truce in Los Angeles decreased

homicides by about 35 percent for the first three months but then doubled in months four

through eleven compared with the pre-truce period Similar findings were reported in Trinidad

and Tobago (Maguire Katz and Wilson 2013) where it was determined that homicides

declined for a brief period of time (again for about three months) but then increased

substantially over the long term (12 months) These results suggested that gang truces may

produce short-term benefits yet result in long-term adverse consequences

The findings of the present case study suggests that some gang truces might last longer than

previously believed While the number of homicides began to slowly increase about 12 months

following the truce in El Salvador the results clearly showed that almost two years following

the truce homicides still remained below those experienced prior to the truce

We found however that the gang truce did not result in a homogenous decline in violence

across municipalities About 61 percent of municipalities experienced a decline in homicides

but the decline in violence varied substantially between municipalities For example while

about 16 percent of these municipalities experienced a 25 percent reduction in homicides a

number of others (37) experienced a 75 percent or greater reduction Furthermore it is

important to note that about 20 percent of municipalities experienced a modest increase in

40

homicides This suggests that the impact of a gang truce might be variable and could be

dependent on contextual factors We examined the possible influence of these factors by

assessing the impact of social structural factors and the presence of gangs on the municipal-

level impact of the gang truce While we found that social structural factors were unrelated to a

decline in homicides municipal-level gang presence was associated with the decline in violence

as a consequence of the gang truce

We examined this issue further by parsing out the relative influence of the number of MS13

and 18th Street gang members on the street and in prison from each municipality As noted

above we hypothesized that municipalities where gang member presence was high regardless

of their presence on the street or in prison would experience a greater reduction in homicides

because of their increased influence in these areas We found however that the relationship

was not as direct as we would have suspected In particular our analyses indicated that

following the truce the number of MS13 and 18th Street gang members on the street in a

municipality was not significantly related to a decline in homicide but the number of

imprisoned MS13 and 18th Street gang members was associated with a significant change in

homicides following the gang truce

Of special note was our finding of the differential impact of the truce based on gang Although

we found that the number of imprisoned MS13 gang members was associated with a significant

decline in homicides following the gang truce we also found that the number of imprisoned

18th Street members was associated with a significant increase in homicides following the truce

These divergent findings might be associated with each gangrsquos organizational structure and its

capacity to regulate member behavior

Much prior research suggests that imprisoned MS13 gang members have substantial influence

over violence in Salvadoran communities15perhaps even more so than formal mechanisms of

15 Instituto Universitario de Opinioacuten Puacuteblica (Iudop) La situacioacuten de la seguridad y la justicia 2009-2014 entre expectativas de cambio mano dura militar y treguas pandilleras Aguilar Jeannette (Coord) [et al] --1ordf ed --San Salvador El Salv Instituto Universitario de Opinioacuten Puacuteblica (Iudop) 2014- 260 p graacuteficos tablas cuadros y figuras 28 cm (Talleres graacuteficos UCA) httpwwwucaedusviudopwp-contentuploadslibro_la_situaciC3B3n_de_la_seguridadpdf

41

social control such as the police and courts16 Law enforcement officials for years have claimed

that MS13 is one of the most organizationally sophisticated street gangs in the Western

Hemisphere17 The gang has been widely characterized as having a highly vertical organizational

structure and strong control over criminal enterprises in gang-controlled neighborhoods and as

being decidedly capable of enforcing rules through discipline MS13 leadership resides in the

nationrsquos prison system The Ranfla (gang leadership) is comprised of thirteen MS13 gang

members (PNC 2011) who direct coordinate and authorize street crime and other gang activity

from prison Our findings lend support to the idea that MS13 is more organized than the typical

street gang and that imprisoned MS13 gang members exhibit strong influence over their fellow

gang members on the street

Our findings also suggest however that the gang truce had a boomerang effect in

municipalities with high numbers of imprisoned 18th Street members implying that 18th Street

might not have as much organizational capacity to regulate violence on the streets as MS13

The truce provided incarcerated MS13 and 18th Street gang leaders an opportunity to

negotiate with high-ranking officials and influential diplomats This may have increased their

legitimacy inside and outside of their gangs It appears that MS13 was able to exert its span of

control over the communities in which they had influence and they were able to deliver on the

terms of the gang truce negotiations In the case of 18th Street however incarcerated gang

members may not have had the same organizational capacity for communicating and carrying

out directives In fact a review of the gang truce indicated that there was a conflict taking place

between two fractions within 18th Street Consequently the organizational structure and

culture of 18th Street might be more diffuse than that of MS13 and its leadership structure

might not be as strong because of the internal fractures within the gang This might further

explain why homicides increased in 18th Street communities The internal fractures within the

18th Street gang may have resulted in intra-gang violent conflict that was largely contained

16 Instituto Universitario de Opinioacuten Puacuteblica (Iudop) La situacioacuten de la seguridad y la justicia 2009-2014 entre expectativas de cambio mano dura militar y treguas pandilleras Aguilar Jeannette (Coord) [et al] --1ordf ed --San Salvador El Salv Instituto Universitario de Opinioacuten Puacuteblica (Iudop) 2014- 260 p graacuteficos tablas cuadros y figuras 28 cm (Talleres graacuteficos UCA) httpwwwucaedusviudopwp-contentuploadslibro_la_situaciC3B3n_de_la_seguridadpdf 17 httpwwwlaprensagraficacomeua-declara-a-ms-grupo-delictivo-transnacional

42

within 18th Street controlled territories Further analysis is needed to examine this specific

issue

While not the primary focus of this case study we controlled for any impact that the Violence

Free Municipalities program might have had on homicide in 11 municipalities during the truce

As noted above the Violence Free Municipalities program served as the second phase of the

gang truce (Cawley 2013) The program was first proposed by the two mediators and designed

by the Technical Committee on Violence and Crime Reduction which included representatives

from OAS MJPS the mediators and the Humanitarian Foundation (CISPES 2013) The program

involved the mayors of each municipality collaborating with gang leaders to design prevention

and intervention resources for gang members and at-risk youth The Minister of Justice and

Security David Mungia Payes announced that his ministry would facilitate $74 million in

funding from the OAS UN and other donors to implement the programing In exchange gang

leaders agreed to end violence and other criminality in the Violence Free Municipalities (CISPES

2013) The negotiators also agreed to further discuss both gangsrsquo demand to repeal of the

2010 law that increased the capacity of the police and prosecutors to crack down on gangs

(Ayala 2014)

Our findings suggested however that the Violence Free Municipalities program was unrelated

to change in homicides in these communities While much additional research is needed to

understand why the program was not effective it might have been because the gangs had

already agreed to a truce and had already reduced violence to the extent that they could

Conversely the demands made by the gangs as part of the Violence Free Municipalities

program might have been more than could be delivered The time and resources required to

implement the programming and the political capital that was required to repeal legislation

might have been much more than could be delivered Future research is needed to examine the

processes and impact of the Violence Free Municipalities program

Regardless the base rate of violence in El Salvador was reset for a period of time suggesting

that perhaps the gang truce substantially altered existing norms of violence Klein (1995)

argued that cycles of gang violence (that is perceived or real changes in gang activity) are

43

typically the consequence of seasonality epochal variation (peaks and valleys in violence) and

illusion (it appears as if there is a real change in violence but there is not) In El Salvador we

appear to have observed a self-directed cycle of normative change wherein incarcerated MS13

gang leaders directed a reduction in violence by actuating their organizational span of control

through the gang truce Some of the most powerful and influential gang members in El Salvador

used their political social and economic capital to promote the truce and articulate new norms

of violence They were able to do this by leveraging their informal social control over the streets

through actual or perceived threats of violence against those who violated the terms of the

truce

However following a change in government leadership and the government subsequently

distancing itself from the gang truce the conditions and capacity of MS13 leaders to intervene

in local violence might have deteriorated and violence began to increase substantially This is in

part because the third parties in the negotiation were no longer able to communicate with

government officials about furtherance of the truce Specifically the mediators were no longer

able to negotiate on the behalf of the government and were no longer permitted entrance to

the prisons where they could negotiate with gang leaders

44

Case Study B Gang Trucemdash The Jamaican Experience

In the latest report by the UN Office of Drugs and Crime Jamaica ranked as the sixth most

violent nation in the world with a homicide rate of about 521 per 100000 residents (UNDOC

2011) Policymakers and researchers have attributed the nationrsquos high level of violence to such

factors as drug trafficking (Klein Day and Harriott 2004) access to illegal firearms (Lemard and

Hemenway 2006) and historical processes that include a legacy of conflict between the

nationrsquos two primary political parties (Sives 2002 Figueroa and Sives 2003 Moser and Shrader

1999) all of which have facilitated the entrenchment of the more powerful gangs in

communities of the urban poor Likewise high levels of income inequality and chronic youth

unemployment (Francis and Lyare 2006) problematic urbanization (Stone 1975) social

marginalization and an emergent subculture of violence (Harriott 2008) have contributed to the

nationrsquos violence problem Whatever the causes violence has had a considerable impact on

Jamaicarsquos social and economic development it has decreased investor confidence (Schwab and

Porter 2008) tourism (Harriott 2007) and access to public services It has also increased the

costs related to the health care system (Mansinghand and Ramphal 1993) the criminal justice

system (Caribbean Human Development Report 2012) and the education system (Moser and

Holland 1997)

Jamaicarsquos homicide problem is closely associated with its gang problem The Jamaican

Constabulary Force (JCF) has estimated that some 272 gangs are active in the nation most in or

near the capital city of Kingston (Harriott 2014) Gang types and their respective historical

patterns of conflict matter in Jamaica as these variations may determine their predisposition or

amenability to lasting rather than opportunistic truces Jamaican gangs include territorially

organized crime groups conflict gangs defense crews who regard themselves as defenders of

their communities (Levy 2009) and other less cohesive more transient territorial groups Some

of the latter are predatory others bond around the taken identities that generate conflicts of

other similar groups18

18 These are generally rooted in subcultural issues such as the demand to be treated with respect or with the indiscretions and self-centered aggressiveness of individual members that may be associated for example with sexual competition for the favors of women

45

Estimates of the gang-related homicide rate in Jamaica vary perhaps because the crime is not

clearly defined as far as attributing a death to a gang Regardless researchers agree that the

proportion of homicides that are gang related has increased substantially For example Harriott

(2003) reported a fourfold increase in the rate of gang-related homicides between 1983 and

1997 Likewise Hill (2013) using official police data found an eightfold increase from 2001 to

2009 reportedly in 2001 only about 64 percent (n=73) of the nationrsquos 887 homicides were

gang related compared with 523 percent (n=879) of its 1682 homicides in 2009

To address the problem Jamaica has initiated traditional law enforcement strategies such as

establishing a specialized gang unit (Sinclair 2004) initiating curfews (Sinclair and Tuner 2005)

declaring states of emergency (Jamaican Observer 2010) and implementing community-

oriented policing (Kolpack 2006) It also attempted legislative reforms to curb election fraud

and electoral-related violence that involved local gangs (Levy 2009) None of these strategies

stemmed the tide of gang violence In 2002 the Minister of National Security established the

Peace Management Initiative (PMI) (Henry 2011) to augment governmental and non-

governmental organizational capacity to settle gang disputes in the community through

intervention-based programming such as ceasefires and gang truces

The current study examines the peace initiative instituted by the PMI in Greater August Town

Our objective was to understand the negotiation processes undertaken with and between

gangs and other stakeholders Among other things we were interested in identifying the actors

involved in the negotiations the goal(s) of the negotiations and the strategies employed to

carry them out Most importantly however we wanted to determine whether the gang truce

resulted in the desired outcome a reduction in the number of homicides in the Greater August

Town area

The Greater August Town (Jamaica) Peace Initiative

Greater August Town is located on the northeastern outskirts of the city of Kingston The low-

income area has high rates of youth unemployment and a history of gang-related violence

(Charles 2004 Levy 2009) Over the last decade the Greater August Town community has

sought improved living conditions and revitalization (Levy 2009 95) The arearsquos inherent

46

resilience has been augmented by nearby intellectual and cultural engines such as the

University of West Indies the University of Technology and University Hospital (Charles 2004

38)

Greater August Town is comprised of the communities of August Town proper (which is

fractured into several locales with gang-given names such as ldquoVietnamrdquo and ldquoOpen Landrdquo and

city government-approved names such as Hermitage Goldsmith Villa (Angola) and Bedward

Gardens These socially defined community divisions and subdivisions are markers for the

territorial boundaries of street gangs and therefore in some instances are lines of potential

conflict Some of those boundaries demarcate areas of Greater August Town that are

predominantly supportive of one or another political party but the boundaries do not always

hold political significance politics is but one element in the conflict geography of the area Like

many communities of the urban poor the Greater August Town area is easily mobilized

politically ndash a reality that is understood and at times exploited by street gangs who politicize

gang ldquowarsrdquo in their efforts to build alliances and to neutralize the police In fact the basic

principle of community mobilization in Jamaica is political patronage and clientelism Access to

resources (eg jobs housing education) for the poorest residents is often determined by the

local political party Thus according to Charles (2004 36)

supporters attach themselves to the political parties to get first preference in the

distribution of scarce resources and over time because they are unemployable they

become dependent on their political party for their economic survival These supporters

will kill anyone who threatens the support base of their political party because they

perceive it as a threat to their daily survival

As a consequence some political supporters invest heavily in the electoral contests and

provoke conflicts that affirm their loyalty to their party in order to secure material benefits

from it Political competition is one conflict fault line in what otherwise is a politically

heterogeneous community Specifically political support in Greater August Town is divided

between the Peoplesrsquo National Party (PNP) and the Jamaica Labor Party (JLP) The PNP receives

strong support from residents in August Town proper the upper region of Goldsmith Villa and

47

Bedward Gardens while the JLP is supported by those living in Hermitage and the lower region

of Goldsmith Villa (Charles 2004 ) The division between the political parties in the area as in

much of Jamaica was believed to be at the root of much of the communityrsquos violence

particularly between 1979 and 1993 Some gangs aligned with the PNP and others with the JLP

this often resulted in political boundaries overlapping with gang turf (Charles 2004)

Pre-truce Violence

Nationally the history of gang violence in the Greater August Town area first appeared as

political violence closely associated with the electoral cycle As in many other urban

communities the problem intensified mdash deeply affecting community life in the period just prior

to the national elections of 1980 mdash then continued cyclically until 1993 Gangs have since

harnessed this legacy in order to establish and maintain community support based on common

political affiliations They have used that support to nullify the efforts of law enforcement to

suppress their illicit activities (Harriott 2008)

The conflict profile of these gangs and of the communities in which they are nested has

changed over the decades From the beginning of their involvement in political violence and

territorial control a form of gerrymandering existed that manipulated the voting population

forcing certain individuals out of a particular constituency and preventing those who remained

in the community from voting for the opposing party This was linked both to the electoral cycle

and to the types of inter-gang conflicts that typically were associated with street gangs

(Figuerou Harriott and Satchell) More recently however much of the crime and violence

perpetrated by gangs has resulted from internal conflicts (eg status management disputes

over womengirls or money individual membersrsquo activities that could attract police pressure)

Internal conflict at times has led to gang fragmentation and new alliances that pull more parties

into the conflict escalating the homicide rate and increasing the sense of insecurity among the

general population (Levy 2012)

The most significant of these internal conflicts resulted from the killing of former Jungle 12

leader Neil Wright by members of his own gang Jungle 12 was the most influential gang in

Greater August Town Before his murder in order to increase the gangrsquos access to illicit

48

opportunities in Kingston Wright had been attempting to extend Jungle 12rsquos influence with a

system of alliances with other gangs and recruitment in Kingston (Harriott 2014) In short his

ambition was to transform Jungle 12 from a neighborhood street gang operating on the

outskirts of Kingston into to a dominant organized crime network that could reach into the

heart of the city In pursuit of this goal Wright recruited members from outside August Town

elevating them in the gang hierarchy above the locals This led to status-related conflicts and

resistance to Wrightrsquos leadership within the gang His murder precipitated a split of Jungle 12

into three factions two of them fled to other neighborhoods within August Town (Angola and

Vietnam) resulting in the formation of new alliances and a new conflict geography that

replaced the former political geography of conflict Wrightrsquos killing and the subsequent demise

of Jungle 12 as the dominant gang in Greater August Town altered the balance of power and

escalated inter-gang violence (Harriott 2014) The post-2005 phase of conflict was

characterized by power symmetry conflict intensification and the spread of conflict

throughout the entire geographic area of Greater August Town

Although their origins are unclear retaliatory killings and other violent incidents progressively

intensified between 2005 and 2008 The violence was episodic retaliations were most often

motivated by suspicions related to personal and geographic connections between warring

gangs As the violence escalated new alliances were formed to enhance power and dominance

which in turn increased the number of gangs and gang members involved in the violence

(Harriott 2014) This eventually attracted national attention and triggered community

mobilization for a gang truce

The Truce-making Process

The Greater August Town gang truce was preceded by frequent intense violence and public

outrage As noted above the violence had escalated in November 2005 when Jungle 12 leader

Neil Wright was killed The defection of a Jungle 12 member to Goldsmith Villa (Angola) caused

infighting within the gang and conflict between it and Goldsmith Villa Just a few months later

Wrightrsquos brother Steve and two others were injured during a turf battle (Martin-Wilkens 2006)

49

Thereafter violence began to occur at regular intervals until January 2007 when the Peace

Management Initiative (PMI) hosted a peace march in the community Two PNP politicians

urged the community to unite A PMI leader declared that the peace march was being held to

ldquodemonstrate to the public that Jungle 12 members are back together and that they want

peacerdquo (Thompson 2007 1)

Although hopeful some residents remained skeptical about the peace march perceiving the

action to be politically motivated In the absence of trustworthy information inter-group

conflicts tended to be interpreted through a politically partisan lens this created obstacles to

isolating the gangs building a consensus for peace and unifying community mobilization The

politically based narratives weakened the communityrsquos leverage for peace as well as the

exposure of the gangs to police action As one resident said ldquoThe election is coming up and

they want[ed] the people to vote for the PNP is one of the main reasons why they have to walk

todayrdquo (Thompson 2007 1) Those who shared such views stayed away from the peace march

Although that widely held myth was not factual it did serve to demoralize and demobilize one

part of the community A local UWI faculty member was articulate in his assessment of the

politics behind the march

August Town violence is not violence of organized crime which is based on drug

trafficking extortion or some other criminal enterprise [but] rather the violence in

August Town is essentially ldquotribalrdquo mdash the Peoplersquos National Party tribe versus the

Jamaica Labour Party tribe [which has been complicated by a splintering within the

PNP tribe] (2007 1)

Indeed it was not the violence of organized crime but neither was it political violence The

individual quoted above neglected to mention that the conflict was between those pro-PNP

splinter groups who were largely comprised of members of Jungle 12 Moreover their ldquopro-

PNP-nessrdquo was unrelated to the conflict there was no factional infighting within the local PNP

organization at that time

Nevertheless following the peace march the gang violence diminished Then in November

2007 a turf war erupted between two gangs from the Greater August Town neighborhoods of

50

Vietnam and River This time as the police stated the gang violence was less about politics and

more about dominance and turf Police were dispatched to perform directed patrols but

whenever they were not present the shootings continued (Mcleod 2007) In April 2008 the

community witnessed local gangs engaging in a five-hour-long street battle that left two killed

and three others wounded It ended only after the police deployed armored trucks The next

month another round of gang violence resulted in five others being killed including a one-year-

old child This resulted in the three members associated with the gang who committed the

homicides being killed in retaliation (Virtue 2008) The local community mobilized against the

violence increasingly cooperating with the police providing more information about the gangs

Subsequently gang members observed a decline in their influence within the community

During the early period characterized by low-intensity conflict the less influential gangs at

times used manipulation of the police as a tactic for suppressing the more influential gangs

This was largely done through strategic release of information Prior to 2005 when Jungle 12

was dominant its membersrsquo illicit activities were constantly reported to police by members of

other gangs as a means of compelling a compromise or settlement of conflicts In practice this

was done by ldquotrading casesrdquo Once a crime had been investigated by the police and suspects

had been charged an opportunity was created for the gangs and other parties to the conflict to

settle the matter by agreeing to drop their cases (typically by ceasing cooperation with police

investigators) This type of ldquoself-helprdquo served to end some of the retaliations but it rested upon

the manipulation of the police (Harriott 2014)

Later in an attempt to quell escalating inter-gang violence the police established buffer zones

between the warring gangs This action resulted in unintended consequences For example

when the police declared a buffer zone between August Town and Hermitage Hermitage took

advantage of the opportunity to attack Angola Some Angola residents accused the police of

turning a blind eye and creating an opportunity for Hermitage to attack their community

Although little reliable information exists about why the police made the deployment the way

that they did it is more likely that the police inadequately assessed the situation (ie mis-

assessed the pattern of alliances and the likely targets of attack) (Harriott 2014) In the areas

affected by this kind of increasing violence community members became angered and lost

51

confidence in the police The error resulted in some parties to the conflict receiving increased

support from their communities and in greater gang-community cohesion (Harriott 2014)

After a brief period the police identified this problem and began to disengage by no longer

providing a buffer between gang controlled areas which in turn allowed still more conflict to

occur between the gangs

As the violence escalated beyond their control police finally responded by applying their own

forms of pressure For example units under the direction of the JCF High Command would

make periodic raids in the community during which they would at times seize weapons and

make mass arrests (Sinclair 2005) However there were also moments when the local police

were very responsive improved their relations with the community and consequently gained

greater access to relevant information Two such moments occurred just prior to and again

immediately after the truce moments during which there was greater freedom of movement

and open collaboration between the community and the police (Harriott 2014)19

The Establishment of the Greater August Town Gang Truce

The Greater August Town gang truce was led by the Jamaican Peace Management Initiative

(PMI) The PMI is a government-funded initiative created for the purpose of working with gang

members to reduce violence Due to community mistrust of the police in 2002 the organization

was established as an alternative organizational mechanism for responding to gang violence

The PMI sought to bridge government and civil society efforts to mediate disputes between

gangs as well as to provide outreach to gang members (Bakrania 2013) While efforts to

institute a gang truce in Greater August Town were led by the PMI a number of other

stakeholders helped to facilitate the truce these included faculty at the University of the West

Indies (UWI) and representatives from the police the faith-based community and the August

Town Sports and Community Development Foundation (Jackson 2008 Levy 2009 also see

Appendix C) The gangs involved in the truce included those from August Town Hermitage

19 The quality of police-community relations largely depended on the style of the local station commander however regardless of the external environment

52

Goldsmith Villa Bedward Gardens and African Gardens Because of its formality as well as its

perceived effectiveness the truce signed on June 24 2008 was regarded by many as the first

of its kind in Jamaica (Levy 2009)

Truce negotiations began early in June 2008 and lasted for about three weeks The gangs

sought to leverage their violence-making capabilities and demanded payment for peace They

asked the third-party negotiators for money ldquoworkrdquo and start-up funds for proposed micro-

businesses (Wilson 2014) Those demands were rejected by the negotiators on the grounds that

the third-party institutions would not buy a peace that was intended to save the lives of those

who were making the demands Moreover if peace was to be purchased then gang conflict

could be used continuously to extract money and other benefits from negotiators The third-

party actors made some demands of their own In some quarters of the community and society

the surrender of guns was viewed as a litmus test of the sincerity of the gangs Consistent with

this the negotiators suggested that the parties to the conflict symbolically hand over one gun

each that suggestion was immediately rejected by the gang leaders These kinds of demands

from the various parties ceased after a time as they all agreed that the truce was to stand on

its own merits (Harriott 2014)

As the truce began to be committed to paper a number of stakeholders expressed concern that

their greatest risk in participating could be the potential for Jungle 12 factions to use the peace

agreement as they had in the past as a tactic to persuade their enemies to let their guard

down Others however recognized that Jungle 12 had now been weakened and that a formal

public peace agreement would be beneficial to the gang and therefore this time would be

different (Harriott 2014)

The gangs held fast to their claim that their weapons were needed for their own protection

because the police were ineffective in responding to violence in their communities (Jamaican

Gleaner 2014) It became a precondition of the truce that the gangs would not be required to

turn in their guns and other weapons (Jackson 2008) The truce agreement did specify

however that ldquoall persons are allowed to move freely across all boundaries regardless of

reputation or affiliation No gun salute or any other shooting is to take place in the community

53

for a period of at least five yearsrdquo (2008 also see Appendix C) The truce agreement and its

conditions were prescribed in a document that was finally signed by all of the major

stakeholders including the gangs (see Appendix C)

Throughout the negotiations each of the gang leaders had attempted numerous times to use

the truce as an opportunity to bargain for money jobs and business support grants Such

demands consistently were rejected by the third-party actors Nonetheless both prior to and

after the truce some efforts were made to create better opportunities for young people

residing in the community UWI for example provided a homework supervision program to

encourage students to further their education and it developed a community-building initiative

to help improve schools and enhance sports programming (Levy 2009) Such programs were

conducted as part of UWIrsquos Township Project in August Town which invested significant

resources in developing the residentsrsquo job-related capabilities and collective self-efficacy

The Greater August Town gang truce was noteworthy for two reasons First the gang truce

received substantial press attention The media were invited to witness the ldquosigningrdquo of the

truce by the gang leaders in the presence of a JCF Deputy Commissioner of Police a PMI board

member and the UWI Principal and two professors of its faculty Second the truce was widely

credited with decreasing violence in Greater August Town and it served as an exemplar to

other communities seeking to replicate its success (Virtue 2008) A number of reports

manuscripts and newspaper articles proclaimed the truce to be a success Bakrania (2013 10)

for example reported that ldquoPMI has been credited with stopping gang wars in August Town

rdquo Levy (2009 94) remarked that the ldquomost interesting outcome of PMI efforts to date was the

Peace Agreement reached in August Town in late 2008rdquo Likewise a government report noted

that ldquothe peace treaty was a pivotal achievement in August Town that has significant potential

for wider application Crime levels dropped markedly in August Town after the signing of the

peace agreement in June 2008rdquo (McLean and Blake-Lobban 2009 78) To this day August Town

celebrates the signing of the truce with an annual celebration with food and music

(Cunningham 2011)

54

Methods

Our evaluation relied on a pre-testpost-test quasi-experimental group design Our

methodology examines the Greater August Town community which is comprised of three

contiguous towns where the gang truce took place (the target area) and the balance of Jamaica

which is comprised of 178 communities (comparison areas) As seen in Exhibit 12 the average

number of residents living in each of the three communities in the target area was not

significantly different than that for the rest of Jamaica about 7776 residents lived in each of

the Greater August Town communities compared with 6468 in the other communities

Likewise communities of Greater August Town were about as densely populated as other

communities (2960 per square kilometer versus 2647 per square kilometer) and the age

range of residents was similar as well However Greater August Town (a) had a significantly

higher proportion of its residents living in poverty (196 vs 158) (b) consumed fewer

resources than other communities and (c) reported significantly more homicides than other

communities prior to the truce (see Exhibit 12)

55

Exhibit 12 Descriptive characteristics of Greater August Town and balance of Jamaica (2007-2011)

Comparison

Area Greater August

Town All areas

Population (mean) 6468 777633 648994

(sd) 720482 353731 715621

Population density (mean) 264719 296033 265238

(sd) 271023 285501 270465

Percent in poverty 1577 1957 1583

(sd) 1036 106 1029

Consumption 15737890 1106939 1566048

(sd) 10713020 205336 1064021

residents under 15 yrs old 2369 2494 2371

(sd) 487 115 484

residents 15-65 yrs old 6840 6901 6841

(sd) 423 29 419

Murder per month (mean) 674 857 677

sd 1928 1409 1920

Total murders 10068 180 10248

n 178 3 181

plt=05

Measures

Two distinct data sets were merged to measure the impact of the Greater August Town truce

First data from the 2011 decennial census provided community-level measures of the social

and economic characteristics of the 181 communities in Jamaica Described in detail below the

community-level data used in the study included population population density gender age

poverty and consumption20 These data were obtained directly from the Statistical Institute of

Jamaica

Second police homicide data from the years 2007 through 2011 were used to construct the

studyrsquos community-level measure of homicide The homicide data were aggregated by month

20 Consumption is an alternative measure of poverty in Jamaica which measures the consumption of food and non-food items

56

and appended to the community-level data The final (merged) data set included 10248

homicides over the 60-month study period These data were obtained from the Jamaica

Constabulary Force (JCF)

The dependent variable examined in the study was constructed from official police homicide

data Once again the homicide data represented the number of officially recognized homicides

in Greater August Town and each of the remaining communities in Jamaica We examined

change by comparing the homicide data 18 months prior to the truce with the homicide data 42

months following the truce More specifically we examined whether there was a change in the

number of homicides in the 30 days following the truce (month 1) as well as whether the truce

had an impact every three months thereafter (ie months 2-5 6-8 9-11 12-14 15-42) and

whether any changes in homicide coincided with changes in homicide in the balance of

observation areas The frequency distribution of our dependent variable is presented in Exhibit

13 It shows that prior to the truce the target area on average experienced significantly more

homicides (1495) than did the comparison areas (920)

57

Exhibit 13 Distribution of homicides in the target and comparison areas

Comparison

Area Target

Area Total

Pre-truce period Mean 920 1495 932

SD 2469 1966 2461

N 241400 5100 246500

Month 1 of truce Mean 741 286 733

SD 1785 496 1772

N 17800 300 18100

Months 2 thru 5 of truce Mean 647 905 652

SD 1821 1249 1812

N 71200 1200 72400

Months 6 thru 8 of truce Mean 577 1236 588

SD 1690 1074 1683

N 53400 900 54300

Months 9 through 11 of truce Mean 718 333 711

SD 2034 733 2019

N 53400 900 54300

Months 12 through 14 of the truce Mean 687 095 678

SD 1519 286 1509

N 53400 900 54300

Months 15 thru 42 of the truce Mean 564 589 564

SD 1683 1042 1674

N 516200 8700 524900

Total Mean 674 857 677

SD 1928 1409 1920

N 1006800 18000 1024800

An illustration of the trends in homicide prior to and following the gang truce are shown in

Exhibit 14 It shows that 30 days following the truce homicides fell in the target and

comparison areas then increased and decreased several times with a general downward slope

in violence over time

58

Exhibit 14 Monthly number of homicides pre-post truce in the target and comparison areas

We also used a number of measures to control for community-level structure from the 2011

decennial census These community-level data included the communityrsquos population

population density (per square kilometer) and community level of consumption Additionally

the census data included measures of the percentage of the population that was female under

15 years old 15 and 65 years old and 65 years old and older as well as a measure of the

percentage of the population living in poverty Principal components analysis was used to

reduce some of these data into a summary measure

Exhibit 15 shows the results of the component loadings One component was extracted that we

designated as socio-economic status (SES) which exhibited high loadings for percent living in

poverty percent under 15 years old percent 15 to 65 years old and consumption Excluded

from the principal components analysis were population and population density Population

was used as our exposure variable and population density was logged to address skewness in

these data

Exhibit 15 Factor loadings from principal components factor analysis

Loading Poverty 78 Consumption -76 under 15 years old 92 between 15 and 65 years old -80

59

Analytic Strategy

In order to test whether the truce had an impact on homicides in the target area andor

whether displacement had occurred in the balance of the study area several analytic

techniques were employed Most of the methods employed the use of the homicide rate as the

dependent variable We explored the data in this way to provide the maximum statistical

power to detect an effect As a check on these methods we also employed a generalized model

to compensate for the non-normality of our outcomes

First focusing only in the target area we performed a simple t-test comparing the homicide

rates before and after the truce (the unit of analysis was a month) however this technique had

limitations the most severe of which was that even if the test were significant it would be

difficult to determine whether the difference was due to the gang truce or a natural change

over time in the outcome Second to address this limitation time series models were

employed whereby the homicide rate for the target area was modeled as a function of time

with truce period indicators included to measure the effect of the truce net of the temporal

trends These models were estimated with ARIMA techniques with a one-month lag auto-

correlated error Third we examined the homicide rate for each town using a panel time series

model In this model the temporal trend for each town was examined with indicators for target

areas and truce periods included The main effects for the truce periods measured the effect of

the truce in the target areas and the moderators of the truce period in the comparison areas

measured displacement effects Finally because the dependent variable coded is not normally

distributed across months we used a negative binomial time series model to estimate the

number of homicides with the population covariate serving as an exposure variable

Findings

The first set of results examines only the target area The first test was a simple t-test

comparing the mean homicide rates before and after the truce periods The result was a mean

difference in the homicide rate of -890 per 100000 with a significant t-statistic of 370 While

60

this result is statistically significant we caution that it may or may not reflect an impact of the

truce To further examine the truce effect in the target area we performed ARIMA regressions

The first model did not include an effect of the temporal trend In this model we again find that

by month 15 the murder rate decreased by about -89 per 100000 (Exhibit 16)

Exhibit 16 Results of basic ARIMA model

Next we employed the ARIMA model again but included a variable (date) to control for the

temporal trends in the data Exhibit 17 shows that when we controlled for temporal trends the

impact of the truce we observed was no longer significant This result indicates that it was not

the truce per se that caused the decline in homicides but instead the decline in homicides was

part of a larger (local and nationwide) decline in homicides

Note The test of the variance against zero is one sided and the two-sided sigma 7877339 5907266 1334 0000 6719536 9035142 L1 1720758 1242218 139 0166 -0713944 415546 ar ARMA _cons 1538436 1904285 808 0000 1165203 1911669 t_15 -8888017 2969383 -299 0003 -147079 -3068134 t_12 -1507749 1820406 -083 0408 -507568 2060181 t_9 -132475 9189969 -144 0149 -3125951 4764507 t_6 -2861008 6335492 -045 0652 -1527835 9556328 t_2 -5166363 4498142 -115 0251 -1398256 3649834 t_1 -1111771 4463623 -025 0803 -9860311 7636769murders_rate murders_rate Coef Std Err z Pgt|z| [95 Conf Interval] OPG

Log likelihood = -2089969 Prob gt chi2 = 00397 Wald chi2(7) = 1472Sample 564 - 623 Number of obs = 60

ARIMA regression

61

Exhibit 17 Results of ARIMA model with control of temporal trends

We next estimated the possible displacement effects of the truce Exhibit 18 presents the

results of these models Examination of the main effects of the truce period does not indicate

any effects and looking at the truceComparison interaction effects we also do not find any

displacement effects Note that these models also controlled for the socio-demographic

characteristics of each community

Note The test of the variance against zero is one sided and the two-sided sigma 7775398 6622814 1174 0000 647735 9073446 L1 1270618 1253989 101 0311 -1187156 3728392 ar ARMA _cons 1480274 1197324 124 0216 -8664378 3826986 date -2318963 2091772 -111 0268 -6418761 1780835 t_15 -3519687 7694691 -005 0964 -1543329 1472935 t_12 -970747 2228007 -044 0663 -5337561 3396067 t_9 -8844212 9402792 -094 0347 -2727335 9584921 t_6 2995697 6576644 005 0964 -1259042 1318956 t_2 -2353533 4923798 -048 0633 -12004 7296933 t_1 -9194625 5462116 -017 0866 -1162501 9786088murders_rate murders_rate Coef Std Err z Pgt|z| [95 Conf Interval] OPG

Log likelihood = -2082031 Prob gt chi2 = 00710 Wald chi2(8) = 1444Sample 564 - 623 Number of obs = 60

ARIMA regression

62

Exhibit 18 Results of Panel (Town) time series model with control of temporal trends

Last we used a random effects negative binomial regression that predicted the homicide rate

with population as an exposure variable and controlled for the socio-demographic

characteristics of each community The results are presented in Exhibit 19 The analysis showed

that time had a negative effect indicating that homicides were decreasing in all areas over the

study period The main effects of the truce (truce = 1 2 ) represented the effects of the

truce in the targeted area and did not show a significant effect for any period following the

gang truce However we did find that the homicide rate significantly increased in the

_cons 6310151 3518738 179 0073 -5864483 1320675lnpopulationdensity 121961 1709831 713 0000 8844891 1554731 ses 1245334 2428629 513 0000 7693312 1721336 date -1276623 0345659 -369 0000 -1954101 -0599144 15Balance 5392222 4128863 131 0192 -2700202 1348464 12Balance 1318516 7903159 167 0095 -2304746 2867507 9Balance 1042658 7925876 132 0188 -5107854 2596101 6Balance -1703928 7919006 -022 0830 -1722489 1381704 2Balance 3498997 7175155 049 0626 -1056405 1756204 1Balance 1056874 1104696 096 0339 -1108291 3222039 truceaugust Balance -4757021 3276618 -145 0147 -1117907 1665032 august 15 -4269255 4284086 -100 0319 -1266591 41274 12 -1262496 7871907 -160 0109 -2805361 2803697 9 -1014276 7881965 -129 0198 -2559113 5305603 6 6563019 786731 008 0934 -1476334 1607595 2 -5034368 7123478 -071 0480 -1899613 8927392 1 -1215158 1095699 -111 0267 -3362688 9323715 truce murders_rate Coef Std Err z Pgt|z| [95 Conf Interval]

Prob gt chi2 = 00000 Wald chi2(16) = 12359 max = 60 avg = 5661878Estimated coefficients = 17 Obs per group min = 48Estimated autocorrelations = 1 Number of groups = 181Estimated covariances = 1 Number of obs = 10248

Correlation common AR(1) coefficient for all panels (02045)Panels homoskedasticCoefficients generalized least squares

Cross-sectional time-series FGLS regression

63

comparison areas in months 12 through 14 following the truce In particular we found a 29

percent increase in the homicide rate in the comparison communities for that period (exp (-

1797 + 2048) = 1285 p-value = 004) Since this effect is only significant at the 005 level

however and given the number of analyses used to examine the data it is possible that we

found this effect by chance alone

64

Exhibit 19 Random Effects Negative Binomial

65

Given these caveats we visualized this model with the following set of marginal predictions as

observed in Exhibit 20 We saw that the targeted area (as illustrated in red) experienced an

immediate decrease in homicide which coincided with an increase in homicides in the balance

of the study area However the target area quickly returned to ldquonormalrdquo and homicides in the

comparison area decreased again During months 9 through 11 following the truce there was a

reduction in homicides in the target area with an associated increase in the comparison area It

is important to point out that the confidence intervals are large and we cannot yield concrete

conclusions from these results However it appears that the truce might have had a temporary

short-lived displacement effect decreasing homicides in the target area but increasing

homicides in the comparison area

Exhibit 20 Predicted change in homicides in the target and comparison areas

Conclusions

From 2000 through 2009 Jamaica experienced a substantial number of homicides many of

which were attributed to gangs in one form or another Traditional law enforcement responses

were repeatedly implemented but until 2010 those had little effect Some policymakers in

Jamaica as well as in other nations throughout the Caribbean and Central America have

recently been experimenting with novel approaches to reducing gang-related violence notably

the implementation of gang truces In Jamaica at least eight gang truces reportedly have been

66

negotiated since 2001 (Levy 2009) The Greater August Town gang truce was thought to have

been one of the more successful and it has served as a model for other communities to use

(2009) Our purpose here has been to identify the actors involved in the negotiations of that

truce the negotiation goals and the implementation methods used and then to examine

empirically the impact of that truce on homicide rates in the targeted community

The 2008 gang truce in August Town was a response to violence that arose when the leader of

one gang was killed creating a power vacuum that other gangs saw as an opportunity to

increase their influence in the community Concomitantly the community as well as the gangs

feared that an absence of formal social control would result in further violence The police

reacted unevenly At some times they engaged in appropriate but heightened levels of

preventive patrol while at other times they purposely provided little or no protection on

occasion they used aggressive tactics that further isolated them from the community The end

result was that there was neither stability nor predictability in the police response and

therefore little trust in the police to address the problem

As the violence further escalated the community mobilized The Jamaican Peace Management

Initiative faculty members from the University of the West Indies (UWI) the Jamaican

Constabulary Force (JFC) and other community-oriented groups joined forces seeking to

reduce the increasing number of homicides by brokering a truce between the gangs Over the

three-week negotiation period the negotiators and the gangs sought terms from one another

The gangs wanted payment ldquoworkrdquo and funds for micro-business development to end the

violence The third-party stakeholders wanted the gangs to disarm actually or symbolically

Neither the gangs nor the stakeholders had substantial leverage nor did they have much to

offer one another in terms of incentives In the end however a truce was agreed upon and all

of the gangsrsquo leaders and several key community stakeholders signed it at a public ceremony

with the media in attendance

At first glance our impact findings appeared to show that the gang truce was an effective

mechanism for reducing violence Bivariate analyses showed a significant decline in homicides

after the truce was implemented This explained the work previously published by

67

policymakers researchers and news reporters Upon further examination of the data however

comparing change in the target and comparison areas and accounting for temporal trends we

found that the decline in homicide was part of a larger nationwide decline in violence and that

the gang truce was not responsible for the decline The only significant effect that we

uncovered was that possibly the homicides were displaced outside the target area for a brief

period of time but then returned to normal

Any one of a number of explanations might be offered for the strategyrsquos lack of effectiveness It

might be that the Jamaican gang leaders at least those in Greater August Town did not have

the organizational capacity to change gang member behavior Much prior research suggests

that in general gangs have limited organizational structure and little formal leadership This

might suggest that gangs do not possess the necessary capacity to regulate their membersrsquo

violence That said gangs in Jamaica including in Greater August Town have been found to be

fairly organizationally sophisticated and to possess strong leadership

In fact in a small number of Jamaican communities gangs have been found to be highly

organized with individual gang leaders being referred to as dons and community leaders The

gang leader in such a community is often found to have substantial control over members and

residents as these communities often turn to the don rather than the police for justice The

don will hold court and punish those who commit crime Punishment can include beatings and

torture as well as execution (Morgensen 2004) Although this level of organizational structure

and sophistication is found only in a small number of Jamaican communities generally the

gangs in Jamaica are believed to have some organizational capacity or at least enough to

reduce violence in communities

Our findings however indicated that prior to 2005 and the death of Neil Wright perhaps only

Jungle 12 could approximate that capacity to discipline members and enforce a truce After the

gangrsquos fragmentation in 2005 Jungle 12 lost much of its organizational capability and

enforcement of the truce was therefore difficult The truce negotiators sought to address the

enforcement issue by proposing a peace council that would involve all parties The proposal

was approved by all key stakeholders still some gang leaders demanded cash payments as a

68

condition for attending council meetings Peace was consistently seen by then as a bargaining

tool rather than as an honest attempt to establish and maintain peace In the end members of

only two gangs were attending the meetings21 and the council soon dissolved

In an effort to replicate the council function UWI sponsored one of the most respected

negotiators a community activist to become a one-person monitoring and intervention

specialist or a ldquoviolence interrupterrdquo His job was to ensure that truce violations did not lead to

a return of the gang wars mdash and there were many violations of the truce For example there

were instances of gang members crossing boundaries and entering the turf of another gang

armed although not initiating conflict behavior that was interpreted by the opposing gangs as

preparation for the next round of ldquowarrdquo or as laying a foundation for a surprise attack that

would exploit the truce for this purpose In the absence of the council these matters were

reported to the violence interrupter who tried to resolve the problems in consultation with the

various gang leaders Often the gang leaders were unresponsive or incapable and therefore the

threatening practices and violence continued Ultimately there were no rules or bodies or

persons who could regulate the violence and there were never any reference points for

compliance The formal truce agreement was an attempt to negotiate and impose such rules

via a collective pressure that would include third parties but it was unsuccessful in doing so

The potential for re-engineering norms related to conflict thus was not realized

Another explanation for the failure of the gang truce might be that it was more a vehicle for

rhetoric rather than for reality The gang leaders insisted that they would sign the truce

agreement only if it were ratified in public with the presence of the media (Jackson 2008 Levy

2009) The leaders might have viewed the process in and of itself as a means of increasing their

reputation and influence within the community and in policymaking circles (and to reduce

mutual distrust) In signing the truce gang leaders publicly pledged to reduce their involvement

in violence thereby calming local residentsrsquo fears They also made public efforts to increase

resources for their communities perhaps in an attempt to portray themselves as ldquoprovidersrdquo to

the community In fact the truce did provide gang leaders with an opportunity to be seen in

21 Interestingly the Jungle 12 factions did not attend any of the peace council meetings

69

public collaborating with important community stakeholders The imagery of the public signing

was of the government (via the PMI) and others approaching the gang to ask them to use their

means of informal social control in the community to reduce violence mdash to accomplish

something that the government could not do on its own As a consequence the process may

have been perceived by gang leaders as a victory because it enhanced the gangsrsquo reputation

with both the government and community

Alternatively from the start the gangs might not have been fully invested in the gang truce

One of the major criticisms of the Greater August Town gang truce was that gangs were not

required to give up their firearms although some believed that this was an unrealistic request

their demand and the demand of many that all guns be turned in immediately was

quite unrealistic given the decades of ingrained gun culture and the continued inability

of the security forces to guarantee protection for any corner against armed rivals It was

obvious to most observers that that kind of situation could not be ended overnight and

that this was a reasonable first step in the process (Levy 2009 63)

The gangs feared that if they were to disarm themselves they would be vulnerable to other

gangs and unable to protect themselves a concern that appears not to have been addressed by

mediators Indeed at times some elements within the community felt somewhat dependent on

the gangs to maintain security If the gangs would have been disarmed and there were no

near-term alternative prospects for any form of social control both the gang and the

community might have faced additional violence as has been observed in the past In the end

the gang truce only called for a reduction in gang violence and did not provide any solutions to

address the larger problems between the gangs nor did it provide the gangs with any tangible

benefits for abiding by the truce

70

Case Study C Gang Trucemdash The Honduran Experience

Introduction

Violence in Honduras is at epidemic levels increasing almost 44 percent over the past five

years In 2012 there were 7172 homicides in Honduras or about 86 homicides per 100000

population (Instituto Universitario de Democracia Paz y Seguridad 2013) making it the most

violent nation in the world (United Nations 2013) Likewise Hondurasrsquo second largest city San

Pedro Sula has the highest municipal level homicide rate in the world with 1290 homicides

(Instituto Universitario de Democracia Paz y Seguridad 2013) or about 174 per 100000

population (United Nations 2013) In comparison the average homicide rate across the globe is

about 62 per 100000 and the average homicide rate in Central America is about 27 per

100000 (United Nations 2013)22

Much of the discussion about the causes of Hondurasrsquo high homicide rate has focused on its

relationship with international drug trafficking routes gangs and conflict between crime

groups and the government and government instability Estimates of gang involvement very

widely but some have suggested that there are between 12000 (Seelke 2012) to 36000

(Ratcliffe et al 2014) gang memberrsquos in Honduras who typically belong to one of two gangs

MS-13 and 18th Street These gangs are said to be less organized than their counter parts in El

Salvador but are said to be just as involved in extortion and intimidation and perhaps more

involved in drug trafficking because of their stronger linkages with Mexican drug cartels

(Wilkinson 2013)

22 An unusual characteristic of the homicide problem in Honduras is the age of victims Typically in the Western Hemisphere homicide victims are aged 15 to 29 In Honduras however those 30 to 44 have the highest rate of violent victimization For example 1 out of 280 males 30 to 44 years old are the victim of homicide compared to 1 out of 360 males 15 to 29 years old (United Nations 2013) These findings by themselves are suggestive of a chronic gang problem (Spergel 1995) Honduras also stands out in the Western Hemisphere in the proportion of its homicides that involve a firearm In 2012 about 84 percent of the homicides involved a firearm (11) The proportion of homicides that involve a firearm appears to be increasing as well In 2008 79 of homicides involved a firearm compared to 81 in 2009 83 in 2010 (United Nations 2013)

71

Over the past decade the nation has responded with ldquoMano Durardquo (ie iron fist or heavy hand)

The new legislation provided the police with more authority to stop search and detain gang

members The new legislation also permitted the courts to sentence gang members to prison

for 12 years for simply being a member of a gang and allowed the courts to sentence

individuals to even longer prison terms for gang related incidents Concomitantly the military

joined the effort to fight gangs by patrolling neighborhoods along side the police While the

public and media strongly supported the shift in national policy toward Mano Dura much of

the evidence suggests that these legislative and policy changes were not effective as the

number of homicides continued to escalate Some suggest that its lack of success was because

gang members who were arrested were released due to of lack of evidence or those who went

to prison if they were not a gang member before entering prison joined a gang Other critics

point out that the heavy handed approach by the government led to loss in the rule of law as

vigilantes engaged in extra-judicial violence against gang members (Seelke 2012) Still others

said that the legislation and policies never really had a chance of working because of the

general lack of effectiveness of the police and courts and the wide spread corruption

throughout the criminal justice system (Zilberg 2011)

As a consequence of the above policymakers and citizens voiced optimism about the possibility

of a truce between gangs after initial results in El Salvador suggested the strategy might be

effective (Villiers-Negroponte 2013) Honduran church leaders and the Organization of

American States (OAS) began to develop a strategy to implement a similar type of truce in

Honduras and the President offered his personal support in their efforts (Arce 2013) In this

case study we examine the processes that lead to the Honduras gang truce and the nationwide

impact of the truce on homicides In the below section we discuss the major stakeholders who

participated in the truce processes leading up to the truce and the establishment of the truce

Key stakeholders

The primary facilitator for the truce process in Honduras was Archbishop Roacutemulo Emiliani who

had earlier served as the Assistant Bishop of the Dioceses of San Pedro Sula and who received

72

support from the Catholic Church to pursue the truce (Bosworth 2013) Prior to the

negotiations he was well known for his work which attempted to establish peace between the

gangs and his advocacy for prison reform and social reintegration programs for gang members

(The Daily Herald 2013) From the onset Monsignor Emiliani proceeded cautiously to ensure

reasonable expectations among the public and policymakers He maintained publically that ldquohe

didnacutet want to be a salesman of false promises about what was going to occur in the future the

things that they do are unpredictable but we expect to have a declaration of reconciliation

principles with societyrdquo (El Mundo 2013) Additionally he wanted to set reasonable

expectations because he knew that it would be a ldquoslow painful and draining processrdquo (El Nuevo

Siglo 2013) and that ldquoWhat is coming is difficult It is not easy It is complicatedrdquo (Castillo

2013)

As in El Salvador the Organization of American States (OAS) played a major role in facilitating

the peace process alongside Monsignor Emiliani Adam Blackwell served as the Secretary of

Multidimensional Security for the OAS and represented Canada on the Honduras Security

Reform Commission (Willcocks 2014) His participation in the mediation process was requested

by Honduran gang members who were in prison They requested that the OAS help broker a

peace agreement with the Honduran government and to help identify resources that would

assist gang members to obtain legitimate jobs (Associated Press 2013) The OAS together with

the Catholic Church served as a ldquobridgerdquo between the executive branch of the government and

the two gangs Additionally two of the mediators (ie Salvadoran Army officer and Police

Chaplain Monsigor Colindres and former Salvadorian congressman Mijango) who helped broker

the truce in El Salvador provided additional support to Monsignor Emiliani and Secretary

Blackwell They traveled to Honduras to present their experiences with the gang truce in El

Salvador and to convey that a gang truce is a promising and legitimate strategy for addressing

gang violence (Associated Press 2013)

Gang leaders of the two primary gangs in Honduras (MS 13 and 18th Street) also participated

extensively in the negotiation process It was stated that they had become weary of the violent

conflict and understood that a truce would be beneficial to the Honduran people (Servellon

73

2013) From the beginning however a number of the critics of the truce argued that Honduran

gangs did not have the capacity to control street level violence They characterized the

Honduran gangs as having less organizational leadership (Bossworth 2013) less control over

turf (LatinNews Daily Report 2013) and being more organizationally fractured (Dudley 2013)

than MS 13 in El Salvador

At the time that discussions about the possibility of a gang truce began President Porfirio Lobo

Sosa was publically supportive of the Catholic Church and OAS negotiating with the gangs

Media reports quoted the President saying I am ldquoprepared to do what ever is necessaryrdquo to

support the mediators (Phillips 2013) ldquoWe have to look for anything thatrsquos an alternative to

violencehellipOn the part of the government we are open to any process that can lower violencerdquo

(Associated Press 2013) and that he had given ldquohis blessing to Emillanirsquos efforts to broker peace

between the gangshelliprdquo (The Daily Herald 2013) However in November 2013 after a general

election the new president Juan Orlando Hernaacutendez through his recently appointed Vice

Minister of Security declared that the government would no longer support the truce process

with the gangs (El Heraldo 2014) Since then the Government of Honduras has not mentioned

the peace process that was initiated in May 2013

Truce making process

It is important to note that prior to the announcement of the gang truce a number of key

stakeholders were somewhat skeptical about its possibility On the one hand some suggested

that a gang truce had been attempted in the past with no success For example one

stakeholder commented to an international media outlet that ldquoEveryone here agrees itrsquos a

positive step forward but people are cautiously optimistic because in 2005 these two gangs

had another peace treaty with each other Now that treaty was very tentative it only lasted

less than two monthsrdquo (Al Jazeera 28 May 2013) On the other hand as noted above other

stakeholders believed that the local gangs did not have enough organizational leadership to

change the behavior of gang members to reduce violence (Bosworth 2013) They argued that

even if gang leaders wanted a gang truce there was no way of enforcing it on the streets

74

Several months prior to the announcement of the gang truce Carlos Mojica Lechuga an 18th

Street Salvadorian gang leader publically stated that representatives of MS13 and 18th Street

in Honduras spoke with several gang truce key stakeholders in El Salvador for the purpose of

replicating the truce in Honduras Tellingly in reflection of the visit Mojica noted that the

advantage of a gang truce is that it formally recognizes Honduran MS13 and 18th Street leaders

as important political persons within the nation He also noted that Honduran gang leaders

have historically been treated poorly and that a gang truce holds the potential for

demonstrating the political power of each of the Honduran gangs (Villiers Negroponte 2013)

The negotiators used different language to describe the early days of the truce Specifically

they mentioned that there had been a consultative process with the gang leaderships so they

were in a process like lighting23 Prior to the truce media sources mentioned that the gang

leaders were offering to stop violence and to not recruit more youth into the gangs24 The gang

leaders also spoke about the ldquopersecutionrdquo they and their family members had suffered during

the previous years highlighting that they had been prohibited social opportunities offered to

others in society (Arce 2013)

Leading up to the negotiations the leadership of both gangs expressed their interest in three

goals 1) lowering violence and crime 2) reconciliation with God society and the government

and 3) helping to improve the social conditions of their communities Although there was not

written documentation on the exact agreement between the parties one MS-13 leader

affirmed that the pact would include all violence (El Comercio 2013) However when talking

about sensitive topics such as extortions which is one of the main sources of income for the

gangs the gang leader said that ldquowould be taken up at a later daterdquo (ABC Internacional 2013)

Leaders of 18th street made similar general statements about ending violence but they were

more specific about their demands One of the 18th Street leaders stated that ldquowhat we want is

23 El Universal 2013 - httpwwweluniversalcominternacional130531obispo-hondureno-descarta-tregua-entre-las-pandillas 24 (Garcia 2013 -httpwwwlaprensahncspmediapoolsitesLaPrensaHondurasSanPedroSulastorycspcid=338546ampsid=276ampfid=98-)

75

to have a dialog with any commission appointed by President Porfirio Lobo and we are sure

that the situation in Honduras will begin to changerdquo (El Nuevo Siglo 2013)

Establishing the gang truce

On May 28th 2013 with public declarations from leaders of both gangs the gang truce was

announced From the beginning of the process the role of the government in the truce was

unclear (La Prensa 2013) Likewise there was little discussion about the exact nature of the

agreement the terms in which gang members would abide and any benefits that would be

made available to those who participated in the truce For example as one stakeholder

indicated the government never decisively considered viable proposals to give the members of

the gangs any opportunities It is important to note that none of the parties signed a formal

commitment and neither MS-13 nor 18th Street signed any type of ceasefire agreement The

gang leaders were in separate locations and were never in direct contact with each other during

the announcement That is both gangs seemingly agreed to the gang truce without ever

talking to each other The ldquopeace processrdquo was publicly announced on May 28 2013 through

ldquojoint but separaterdquo declarations made by the leaders from both gangs imprisoned in the San

Pedro Sula prison (National Penitentiary SPS) (The Daily Herald 2013) The national and

international media widely covered the declarations

The MS-13 leaders said they would not commit any more homicides or any other types of

crimes They ensured that this was an ldquoimmediaterdquo order and would be effective throughout

the country They emphasized ldquoall of the boys know what they have to do starting todayrdquo

(Pachico 2013) The leadership of 18th Street declared that they would stop violence and other

criminal activities but also indicated that the government would have to ldquolisten to themrdquo

Little research has examined whether the gang truce in Honduras ever impacted violence in the

nation Instead anecdotes have been used to portray its effectiveness One facilitator for

example indicated ldquoin Honduras the dialog with the gangs has been positive however the sad

thing in Honduras is that the two main gangs have not accepted a truce between them they

just haven`t accepted it as yetrdquo (La Prensa 2013) despite the fact that ndash in their own words ndash

76

ldquothey do want to hold a dialog with society with the government and with the policerdquo

Conversely gang leaders declared that the truce had been effective For example a leader of

18th Street noted that ldquohellipIt has already done its part telling members in the areas the gang

controls to stop the violence and crimehellip[estimating] crime had already dropped 80 percent in

those areasrdquo (Associated Press June 17 2013) Similarly a member of MS13 estimated that

violence in MS13 controlled areas declined by 45 percent (Associated Press June 17 2013) As

a symbolic gesture of the impact of the truce MS13 leaders also noted that as a gesture of good

will they made and delivered 60 beds for a nursery home in San Pedro Sula (Associated Press

June 17 2013)

Methods

For the present case study we used a pre-test post-test longitudinal quasi experimental design

Data from the 2001 Honduran Census was obtained from the National Institute of Statistics

(INE) These data provided municipal level measures of number of residents population density

per kilometer percent of population who moved in from another municipality ethnicity

percent urban number of residents immigrated to the United States percent female headed

households percent unemployed age composition income percent of households rented and

education level In addition population projections for the years 2005 2010 and 2014 were

also obtained from the National Institute of Statistics (INE) The population levels for the total

population as well as the percentage of residents in a municipality that rural and percentage of

residents who are female were linearly interpolated for the intervening years Examination of

the observed levels indicated that growth was linear overall and so we feel confident that our

linear interpolations are good approximations Second we used municipal level homicide data

by month and year for the period May 2012 through July 2014 These data were provided by

the Honduran National Police through the Honduran US Embassy Both datasets were

merged for the present analysis

Measures

77

The dependent variable for the Honduran case study is the monthly homicide rate which was

calculated by dividing the number of homicides in each municipality by its population and

multiplying this figure by 100000 We examined change by comparing the homicide data 13

months prior to the gang truce to the homicide data 14 months following the truce in each of

the nationrsquos 298 municipalities As presented in Exhibit 21 there were a total of 7910 homicides

over the study period with each municipality averaging 183 homicides (sd=837) A trend

analysis showing the monthly number of homicides on a national level prior to and following

the gang truce is presented in Exhibit 22 It shows that nationally the homicide rate gradually

declined over the study period

Exhibit 21 Summary Statistics

Pre-Truce (n=3809) Post-Truce (n=4101) Total (n=7910)

Mean SD Mean SD Mean SD

Homicide 200 900 166 773 183 837

Population 2874492 8674659 2939557 8870267 2908226 8776124

Homicide rate 521 953 413 814 465 885

Density 9926 10878 9928 10878 9927 10877

Born in the same municipality 8211 1286 8211 1286 8211 1286

Other population group 8545 2390 8545 239 8545 2390

Percent rural 7926 2732 7899 2762 7912 2748

Living in another country 331 650 331 65 331 650

Socioeconomic status 005 097 -002 102 001 100

Exhibit 22 Homicide rate in Honduras by month

78

We used census data to control for several community-level structural factors Principal

components analysis was used to reduce some of the data into a summary measure Exhibit 23

presents the results of the component loadings One component was extracted that we labeled

socio-economic status (SES) This component exhibited high loadings for percent female

headed household percent unemployment and percent completing primary education

Population was used as our exposure variable and population density (per 1 km) residential

stability percentage of the population that is rural percentage of the population that is

indigenous (verify) and the number of residents immigrated to the United States served as our

control variables Population density and outmigration were logged to address skewness in

these two measures

Exhibit 23 Factor Analysis for Socioeconomic Status

Eigenvalue 236093

Variables Factor Loadings

Female-Headed Households 07977

Unemployed 09269

Primary Education 09303

Findings

The results of our t-test analysis are shown in Exhibit 24 It shows that there was a significant

decline in the homicide rate at the municipal level before and after the gang truce The

homicide rate prior to the truce was 697 per 100000 population and following the truce the

homicide rate was 566 homicides per 100000

79

Exhibit 24 T-test on National Data by Month (n= 27 Months)

Homicide Rate Pre Truce Post Truce Change

Mean 6972346 5663592 -1308754

SE 01677361 01151668 02009224

plt 001

Next we examined the effect of the truce through ARIMA regressions The first model in Exhibit

25 does not include an effect of the temporal trend In this model we again find that homicides

declined significantly in the period following the truce We then employed the ARIMA model

again but this time included a variable (month) to control for the temporal trends in the data

Model 2 in Exhibit 25 shows that when we controlled for temporal trends the impact of the

truce we observed was no longer significant Our findings suggest that homicide did not decline

as a consequence of the gang truce but instead the decline in homicides was part of a longer

term historical trend that was independent of the truce

Exhibit 25 Results from the ARIMA Models (n= 27 Months)

Homicide Rate Model 1 Model 2

Effect SE Sig Effect SE Sig

Truce -1309 0350 -0906 0595

m2 -0623 0481

m3 -0153 1447

m4 -0074 0607

m5 -0104 0450

m6 -0050 0466

m7 -0002 0516

m8 0003 0404

m9 0024 1633

m10 -0396 0522

m11 -0754 2424

m12 0837 0366

Time -0030 0047

Intercept 6973 0254 7258 0472

Autocorrelation Coefficient 0010 0408 0111 0249

plt 001

80

Last we used a fixed effects multi-level negative binomial regression to predict the homicide

rate with population as an exposure variable and controlled for the socio-demographic

characteristics of each community The results are presented in Exhibit 26 The only significant

variable in the analysis was the impact of municipal level population stability Specifically we

found that homicide rates increased in areas where residential mobility was high Once again

our analysis showed that time had a negative effect indicating that homicides were decreasing

in all areas over the study period The gang truce itself was unrelated to the decline in

homicides

Exhibit 26 Results from the Multilevel Negative Binomial Models (n= 27 Months)

Homicide Rate Model 1 Model 2

Effect (SE) Sig Effect (SE) Sig

Fixed Effects

Homicide Rate 0004 0004

(0002)

Truce -0119 -0129

(0073) (0072)

Time -0008 -0007

(0005) (0005)

Month 1 0024 0024

(0059) (0059)

Month 2 -0107 -0107

(0062) (0062)

Month 3 -0041 -0041

(0063) (0063)

Month 4 -0033 -0033

(0065) (0065)

Month 5 -0002 -0002

(0067) (0067)

Month 6 (reference)

Month 7 0027 0027

(0051) (0051)

Month 8 0023 0023

(0057) (0057)

Month 9 0036 0036

81

(0056) (0056)

Month 10 -0051 -0051

(0057) (0057)

Month 11 -0031 -0031

(0058) (0058)

Month 12 0187 0187

(0056) (0056)

Born in the same municipality -0013

(0003)

Pecent rural -0003

(0002)

Socioeconomic status 0029

(0044)

Percent dominant population 0001

(0002)

Density (ln) -0010

(0050)

Living in another country (ln) 0001

(0061)

Intercept -10042 -10051

(0061) (0058)

ln(Alpha) -2368 -2367

(0148) (0148)

Random Effects

Var(Truce Effect) 0066 0066

(0024) (0025)

Var(Intercept) 0388 0309

(0049) (0042)

Cov(truce effect intercept) -0024 -0006

(0027) (0025)

plt 001

82

Conclusions

Violent crime in Honduras is widespread With a homicide rate of about 86 per 100000

population Honduras is one of the most violent nations in the world (United Nations 2013)

This compared to an average homicide rate around the world of 62 per 100000 people and

about 27 per 100000 in Central America (United Nations 2013) Given the high rate of violence

in Honduras it is clear that new and innovative ways of reducing violence should be considered

This case study gave an overview of the implementation and impact of the gang truce

experience in Honduras in 2013

The goal of the truce was to significantly reduce the number of homicides In order to assess

the impact of the gang truce in Honduras a pre-test post-test longitudinal quasi-experimental

design was used Census and homicide data were merged at the municipal level to asses

whether the truce had an impact on homicides controlling for population characteristics and

the natural trend in violent crime

Overall the findings suggest that while the homicide rate in Honduras was on a slight

downward trend the gang truce itself was unrelated to any homicide reductions That is the

gang truce had no measurable impact on homicides in Honduras Given the drastic reductions

achieved in El Salvador and the fact that Hondurasrsquo gang truce was a replication of El

Salvadorrsquos the following discussion will examine two important differences between the two

countries gang truces First the implementation of the truce in Honduras did not appear to

obtain trust between all parties involved and did not achieve any notable short-term

deliverables That is the implementation was not robust Second some suggest that the gangs

in Honduras do not have the organizational sophistication to be able to control their members

on the streets rendering them incapable of carrying out any truce agreements

First the implementation of the truce in Honduras did not appear robust The communication

between the parties was weak and none of the parties completed any significant actions as part

of the truce The Catholic Church and OAS served as the bridge between the two main gangs in

Honduras MS13 and 18th Street and the executive branch of the government The ldquopeace

83

processrdquo as it was called in Honduras was initiated in May 2013 After the general election in

November 2013 the newly appointed administration declared that it would no longer support

the truce process with the gangs As a result the truce was short-lived and the governmental

support for the effort shifted with the change in leadership

Though the negotiations were largely based on the good will of a well-respected negotiator

few tangible incentives were offered during the process The negotiations from the beginning

involved discussions about large scale social programming Given the short time frame these

goals in hindsight were unrealistic Naturally there was little trust between the parties at the

beginning of the negotiations and without any quick tangible deliverables from either side the

truce never really materialized

The second challenge to implementing a successful gang truce in Honduras might have been

related to the nature of the gangs themselves It was unclear whether the gangs possessed the

level of cohesion and hierarchical leadership required to make some measures feasible If the

gangs do not have the organizational capacity to control their members on the streets any

agreements that come from the negotiations would be difficult to implement In the earlier

case study of the Salvadorian experience it was suggested that the successes in that country

were related to one of the gangrsquos organizational capacity to impose control of its members The

gangs in Honduras might be different There is at least some evidence that MS 13 in Honduras

might have less organizational leadership (Bossworth 2013) less control over turf (Latin News

Daily Report 2013) and have been more organizationally fractured (Dudley 2013) than their

counterparts in El Salvador The gangrsquos ability to operate as an efficient organization can greatly

impact the outcome of the truce process In general we know that gangs do not have very high

levels of organizational sophistication (Decker Katz and Webb 2008 Decker Bynum Weisel

1998) It might be that gangs in Honduras are more the norm in terms of organizational

capacity

In summary the 2012 truce negotiations in Honduras did not produce any measurable

reductions in the homicide rate The gangs wanted to speak to Honduran society and they even

84

preferred to speak with the governmental authorities but they never interacted with either

The negotiation process seemed to end as quickly as it started The gangs did not deliver with

lower rates of violence and the government did not provide social programs It would be safe to

say that a robust gang truce did not materialize in Honduras The implementation of the truce

seemed to struggle for two primary reasons First the mediators were not able to accomplish

any quick wins to build trust between the parties involved The commitment level on all sides

was not clear throughout the process Second it is not clear whether the gangs in Honduras

have the organizational capacity to control members on the street as would be required to

carry out an effective gang truce In short the 2013 gang truce in Honduras was unsuccessful

85

Conclusions Policy Implications and Recommendations

The purpose of this report was to systematically examine the effectiveness of gang truces Gang

truces have been widely implemented but rarely evaluated Of those gang truces that have

been evaluated little attention has been given to why and how they came into existence In

this report we reviewed prior research on gang truces and presented case studies of gang

truces implemented in El Salvador Jamaica and Honduras for the purpose of understanding

the negotiation processes undertaken with and between gangs and other stakeholders We

were interested in identifying the actors involved in the negotiations the goal(s) of the

negotiations and the strategies employed to carry them out Most importantly however we

wanted to determine whether the gang truce resulted in a reduction in the number of

homicides Each case study offers lessons learned that are unique to their particular

circumstances and when considered together provide direction to policymakers on the

benefits and risks of implementing gang truces

The case studies presented here constitute the most comprehensive evaluations of gang truces

to date Existing documents were used to collect information about the processes associated

with each gang truce Many of these documents included such items as peer reviewed articles

books and reports The majority of these documents were collected over the Internet

requesting documents from those close to the truce and searching library databases Related

the case studies made use of articles obtained from local newspapers The newspaper articles

were not only intended to provide a historical record of the development of each gang truce

but also to provide additional insight into the various external forces that might have impacted

the gang negotiations Because the newspaper serves as a forum for the community to speak

about its concerns newspaper articles also provided a rich source of data on how those in the

community felt about the gang truce Accordingly the newspaper articles offered a different

view of the problem and offered different opinions as to how a gang truce should or should not

be implemented We also conducted a small number of in-depth qualitative interviews with

key informants These data were collected to supplement existing documents and to clarify

issues associated with the negotiation processes This included but was not limited to

86

questions pertaining to identifying the actors involved in the negotiations the goal(s) of the

negotiations and strategies employed to carry out negotiations The interviews were intended

to obtain information from those who possessed first hand knowledge about the gang truce in

each nation

We examined the impact of each truce using official data We first performed a simple t-test

comparing the homicide rates before and after the truce However as discussed above this

technique has limitations The most severe of which is that even if the test was significant it

would be difficult to determine whether the difference was due to the gang truce or a natural

change over time in the outcome We addressed this limitation by using time series models

whereby the homicide rate for the community was modeled as a function of time with truce

period indicators included to measure the effect of the truce net of the temporal trends These

models were estimated with ARIMA techniques Supplemental models were also employed to

examine and control for factors other than the truce that might have impacted homicide over

the study period

Summary of Findings Related to the Implementation of a Gang Truce in the Three Sites

We found that the implementation of gang truces have a number of common characteristics

The first is that in each case a community was experiencing an uncharacteristically high number

of gang related homicides over a fairly lengthy period of time The continued high level of

violence in each case resulted in the community placing strong pressure on the government in

general and the justice system in particular to respond to the problem quickly and effectively

In each case they had first attempted to control gang violence through suppression oriented

strategies and these strategies were found to be ineffective over the intermediate and long

term In turn each communityrsquos inability to exercise traditional informal and formal social

control to decrease levels of violence became self evident to the public and government This

resulted in both the state and community to seek (or participate in) an alternative strategy in

which negotiators would formally andor informally work with gang leaders to establish a truce

that would reduce gang homicide

87

Key stakeholders involved in the negotiation and establishment of each gang truce were fairly

similar In each of the cases examined the gang leaders of the largest and most violently

involved gangs were willing to consider participating in negotiations that could lead to a truce

In each of these cases it was clear that the gangs not only sought to collaborate with the

negotiators for the purpose of reducing violence but perhaps more importantly were seeking a

means in which to gain greater more positive recognition in the community and to reap some

form of benefit to themselves their members and possibly their community In each case

while not always formally involved government officials were at a minimum made aware of

negotiations and in some cases solicited the assistance of third partyrsquos to broker an agreement

between stakeholders In each case it was at least implicitly understood that the government

would ldquolistenrdquo to the gang leaderrsquos expectations of the government and what theymdashthe gang

leaders--had to offer in exchange We found that when the government was no longer willing

to ldquolistenrdquo to or collaborate with negotiators the truce processes ended abruptly Negotiators

were typically comprised of a very small group (ie 2-3) of individuals who were perceived to

be ldquohonest brokersrdquo In El Salvador and Honduras this included a high ranking Catholic Church

official a leader from an international diplomatic organization (ie OAS) and other neutral

parties In Jamaica this included a quasi-governmental organization that had been established

for the purpose of brokering negotiations between gangs to reduce violence and the local

university which had access to staff who were perceived to be neutral but had an interest in

reducing violence due to its proximity to the university

The strategies used to execute each gang truce were similar but yet importantly different They

were similar in that each involved a team of negotiators working to identify common goals to

be achieved and identifying tangibles that could be delivered to the gang leaders gang

members and their community in exchange for the gang achieving their stated goals They

were different however in terms of the structure of the delivery of each parties promise to the

other In Honduras and Jamaica it appears that gang leaders committed to reducing gang

violence in exchange for general promises made by the negotiators for example that

substantial public works programs would be implemented for the goal of reducing

unemployment among gang members and the community In both of these cases it required

88

the government to develop and deploy large scale social programming in a very quick period of

timemdashsomething that neither government had a strong record of demonstrating In El Salvador

negotiators employed a strategy of the gang leaders promising to deliver immediate changes in

gang member behavior for immediate administratively natured changes by the government

For example in exchange for a reduction in gang violence the government agreed to relocate

imprisoned gang leaders to less restrictive prisons and provide them some privileges Following

the successful execution of the first part of the truce which resulted in near term success for

both parties they began to negotiate issues that would take a longer period of time for the

gangs and government to deliver Our findings suggested that some promised deliverables need

to be easily and quickly delivered early in the process so that trust increases between both

parties Stakeholders only have a brief period of time to provide promised benefits before trust

is lost and that tangible benefits need to be delivered in weeks or months not years

Summary of Findings Related to the Impact of Gang Truces in the Three Sites

Our findings suggest that El Salvadorrsquos gang truce had a significant and dramatic impact on the

homicide rate The mean number of monthly homicides declined from about 354 prior to the

truce to about 218 following the truce for a net decrease of about 136 homicides per month

Our forecast showed that between March 2012 and June 2014 the truce had saved about 5501

lives However there was no significant change between the pre-truce and post-truce periods

in the number of thefts extortions robberies rapes and auto theftsrobberies We also found

that the gang truce did not result in a homogenous decline in violence across municipalities

About 61 percent of municipalities experienced a decline in homicides but the decline in

violence varied substantially between municipalities We examined this issue further by parsing

out the relative influence of the number of MS13 and 18th Street gang members on the street

and in prison from each municipality Our analyses indicated that following the truce the

number of MS13 and 18th Street gang members on the street in a municipality was not

significantly related to a decline in homicide but the number of imprisoned MS13 and 18th

Street gang members was associated with a significant change in homicides following the gang

truce In particular the number of imprisoned MS13 gang members was associated with a

89

significant decline in homicides following the gang truce and the number of imprisoned 18th

Street members was associated with a significant increase in homicides following the truce

In Jamaica our initial findings showed that the gang truce might be an effective mechanism for

reducing violence Bivariate analyses showed a significant decline in homicides immediately

after the truce was implemented This explains the work previously published by policymakers

researchers and news reporters Upon further examination of the data however comparing

change in the target and comparison areas and accounting for temporal trends we found that

the decline in homicide was part of a larger nationwide decline in violence and that the gang

truce was not responsible for the decline The only significant effect that we uncovered was the

possibility that homicides were displaced outside the target area for a brief period of time but

then returned to normal

Our findings from Honduras told a similar story as Jamaica Initial analysis showed that the

number of homicides on average declined across municipalities following the gang truce

Specifically the mean number of homicides declined by 13 per 100000 population with 687

homicides per 100000 population on average occurring in each municipality prior to the truce

and 566 homicides per 100000 population on average occurring in each municipality after

the truce However after we examined the effect of the truce through the ARIMA model and

included a variable (month) to control for the temporal trends in the data the impact of the

truce we observed in our bivariate analysis was no longer significant Our findings as in

Jamaica suggested that the decline in homicides was not the consequence of the gang truce

but instead the decline in homicides was part of a long term decline in homicides due to

exogenous factors

The Potential Benefits and Consequences of a Gang Truce

Over the last several years there have been a number of naturally occurring experiments

involving gang truces in a variety of nations in various regions of the world Findings from

evaluations of gang truces are mixed As noted above in El Salvador the gang truce could be

characterized as highly effective at least for the two years following the truce It is worth

mentioning that even after the truce breakup homicides rates while above truce levels

90

continue slightly below pre-truce levels In Jamaica and Honduras the gang truce had no impact

on violence In Los Angeles and Trinidad there was evidence that violence decreased for at

least ninety days but then increased substantially beyond those rates observed prior to the

gang truce (see the introduction section of this report for this discussion) As a consequence it

appears that the potential for long term consequences might out weigh the potential for short

term benefits Only one study site(El Salvador) demonstrated a truce having a substantial and

long term impact on violence Others conversely demonstrated the truce had no impact or

increased violence over the long term In fact a number of scholars have noted that gang truces

are likely to result in a boomerang effect with gang violence increasing over the long run

because of enhanced cohesion within the gang (Klein 1995) Maguire (2013) notes that when

government officials negotiate a truce with gangs they might ldquoinadvertently be acknowledging

gangs as legitimate social entitiesrdquo (p 11) This in itself might increase cohesion among gangs

which has been found to be associated with increased levels of criminality (Decker et al 2008

Klein 1971 Maguire 2013)

Further research is needed to examine how gang truces might impact group cohesion and if it

does whether this in turn results in greater violence Gang truces convey the well-intentioned

image that violence has been addressed and policymakers are doing something about the

problem but researchers need to better understand the probability of a gang truce reducing

violence increasing violence or having no impact This will better position policymakers to

understand the relative risks associated with these types of interventions

Our findings also suggest that while gang truces could be an effective intervention in areas

where gangs are highly structured and organized such as El Salvador they could be counter-

productive in areas where gangs are not as structured and organized Because the vast majority

of street gangs are not well organized (Klein 1995 Spergel 1995) the utility of a gang truce in

reducing violence might be limited Our findings coupled with prior research suggest that gang

interventions need to be tailored to the nature of the gang and its members or it risks

increasing gang violence

Final Thoughts

91

Our analysis suggests that gang truces should only be used as a means of last resort and then

only under certain conditions Given the risks associated with a gang truce communities with

high levels or at least modest levels of formal social control should rely on other more

promising gang control strategies Only when the state has limited or greatly reduced capacity

for social control should a truce be considered Concomitantly a gang truce should only be

considered when a community is experiencing a substantial amounts of gang violence

Communities that are experiencing minimal to modest amount of gang violence may have more

to lose from the establishment of a gang truce than they have to gain Additionally our

findings suggest that a gang truce might only be feasible when gangs are sufficiently organized

to the extent that they have the capacity to regulate memberrsquos behavior In other words gang

leaders must have the ability to reduce their memberrsquos involvement in violence for a gang truce

to work Our findings more concisely suggest that gang truces should only be considered when

there are a great number of gang homicides the state has limited capacity to address the

problem and gang leaders have enough informal social control over their members that they

themselves can substantially control the levels of violence in their community by regulating

their memberrsquos behavior

Recommendations

Gang truces are conjunctural strategies States who suffer from gang-related violence must

establish permanent public policies for crime control and prevention A government that

considers implementing a gang truce should be aware that it cannot become the center

strategy of its public policy for citizen safety

Gang truces should only be used as a means of last resort and then only under certain

conditions Stakeholders must determine whether a process of dialog or negotiation with gangs

is legal ethical and feasible

Stakeholders must anticipate demands that are likely to arise and their response options Some

demands may be easily met such as improved prison conditions Others are much more

difficult and amorphous such as community development through more integrated violence

92

prevention programs (such as those implemented by SolucionES in El Salvador) local economic

development programs or economic reinsertion of ex-gang members

Stakeholders should incorporate immediately achievable and demonstrable deliverables Long-

term goals and promises are unlikely to create the trust needed to sustain a gang truce

Stakeholders must first determine the position in which they are negotiating the incentives that

are possible to deliver and the boundaries and limits they face Gangs are mostly likely to trust

representatives from NGOs community-based organizations and members of the faith based

community as brokers because they are considered more reliably neutral advocates for peace

They need to understand the capacity of the government to deliver promises in a timely

manner

Governments have to make a choice about the visibility and transparency of its participation

This decision needs to be made in the context of the national and local laws the publicrsquos

expectations of transparency and patterns of practices of the past

Governments must be strategic in their support for truce initiatives Some donor funded

programs run by the government prohibit gang member participation and if the government

does not receive approval from the donor it may risk the donor withdrawing its sponsorship of

the program

Governments must ensure an inter-institutional coordination for the management of truces to

avoid the responsibility to be of a single government institution It is necessary to generate or

collect reliable and pertinent data that can be used to analyze and assess the process

It is necessary to implement an effective monitoring system of the truce process similar to that

used in the full report as it can help parties understand what is working and who is delivering

on their promises More specifically the monitoring and truce management system should be

able to identify truce violations and be prepared to respond through the use of legal and

effective practices if stakeholders do not comply

93

Finally it is necessary to develop evaluations of gang truces and monitoring programs and

support violence prevention activities local economic development activities and pilot

programs to support the reinsertion of ex-gang members into society Clearly national

governments municipal governments NGOs and community-based organizations need

increased capacity and resources to discourage the growth of gangs among at-risk youth It is

therefore increasingly important to create economic opportunities for gang members willing to

leave the gangs and find other legal employment Developing and sustaining those

opportunities in nations with high incidences of poverty will require significant international

funding

94

References

Anderson Elijah 1999 Code of the Street New York Norton

mdash 1998 ldquoThe Social Ecology of Youth Violencerdquo In Crime and Justice vol 24 edited by M

Tonry 65-104

mdash 1990 Street Wise Chicago University of Chicago Press

Arce A (2013 May 28) Honduras gangs declare truce ask talks with govt Associated Press

The Big Story Retrieved from

httpbigstoryaporgarticlehonduras-gangs-declare-truce-seek-talk-govt

Archibold Randal 2012 ldquoGangsrsquo Truce Buys El Salvador a Tenuous Peacerdquo The New York Times

Originally published online on August 27 2012

Axelrod R (1986) ldquoAn Evolutionary Approach to Normsrdquo American political science review

80(04) 1095-1111

Ayala Edgard 2012 Gangs Back Plan for Violence-Free Districts in El Salvador

httpwwwipsnewsnet201212gangs-back-plan-for-violence-free-districts-in-el-

salvador Found on November 4 2014

Bakrania Shivit 2013 Policy Responses to Criminal Violence in Latin America and the

Caribbean GSDRC Helpdesk Research Report 934 Birmingham UK GSDRC University

of Birmingham

Bargent James 2013 ldquoRise in Disappearances Feeds Doubts Over El Salvador Trucerdquo August 1

2013 Found at httpwwwinsightcrimeorgnews-briefsrise-in-disappearances-in-el-

salvador-feeds-doubts-over-truce on September 1 2014

Battin Sara R Karl G Hill Robert D Abbot Richard F Catalano and J David Hawkins 1998

ldquoThe Contribution of Gang Membership to Delinquency Beyond Delinquent Friendsrdquo

Criminology 36 93-115

Block Richard 2000 ldquoGang Activity and Overall Levels Of Crimerdquo Journal of Quantitative

Criminology 16 (3) 369-83

Table of Contents

Page Executive Summary i

Section 1 Introduction 1

Section 2 Gang Truce - The Salvadorian Experience 11

Section 3 Gang Truce ndash The Jamaican Experience 44

Section 4 Gang Truce ndash The Honduran Experience 70

Section 5 Conclusions Policy Implications and Recommendations 85

References 94

Appendix A Systematic Review of the Literature 107

Appendix B Truce Related Media Statements in El Salvador 111

Appendix C August Town Five Year Peace Agreement 121

Appendix D Time Line El Salvador Gang Truce 123

Appendix E Social Violence Prevention Programming 124

i

EXECUTIVE SUMMARY

Background

While there is much literature describing the assumptions issues and effectiveness of crime

suppression (eg Decker 2003 Decker and Reed 2002 Katz and Webb 2006 McCorkle and

Miethe 2002) and prevention strategies (Esbensen and Osgood 1997) much less attention has

been paid to gang intervention programming particularly gang truces Little is known about

how often gang truces occur what conditions give rise to them the role of third parties in

brokering them their transformative effects and their effectiveness In this policy brief

sponsored by SolucionES1 and conducted by FUNDE a member of the SolucionES Alliance with

Arizona State University we systematically evaluate gang truces including reviewing prior

research and presenting evidence on the effectiveness of gang truces that have been

implemented in El Salvador Honduras and Jamaica for the purpose of identifying lessons

learned should other governments or donors wish to support gang truces in these or other

countries

Summary of Findings

We found that the truce in El Salvador resulted in a reduction in homicides that was not the

product of other trends or temporal factors Thus the truce in this country had a short-term

effect in reducing the lethality of violent crime though not necessarily in other dimensions of

insecurity By contrast the truces in Jamaica and Honduras resulted in no impact on violence

The Jamaican and Honduran experiences therefore mirror the results of prior gang truces that

1 SolucionES is multifaceted violence prevention program being implemented by an Alliance of five leading Salvadoran non-profit organizations who have come together to prevent crime and violence The Alliance members are Fundacioacuten Nacional para el Desarrollo (FUNDE) Fundacioacuten Salvadorentildea para la Salud y el Desarrollo Humano (FUSAL) Fundacioacuten Crisaacutelida (known locally as Glasswing) Fundacioacuten Salvadorentildea para el Desarrollo Econoacutemico y Social (FUSADES) and Fundacioacuten Empresarial para el Desarrollo Educativo (FEPADE) with partial funding from USAIDEl Salvador Together these five organizations have widely-recognized expertise in education health community development economic development research and youth leadership they are bringing their combined synergy and strengths to prevent crime and violence in El Salvador SolucionES is using a three-pronged strategy to prevent crime and violence in El Salvador 1) Strengthen municipal crime and violence prevention capabilities and actions 2) Increase social investment by the private sector to prevent crime and violence and 3) Research publish and disseminate findings to inform decision-making on crime and violence prevention This is one of 10 planned policy-oriented studies The full study is available on request

ii

have been studied to various extents including those in Los Angeles and Trinidad and Tobago

Important differences in how the various truces were negotiated may explain the different

results and one important difference ndash the ability of government and non-gang community

stakeholders to promise and immediately produce measurable deliverables ndash appear to be

especially significant

Prior Evaluations of Gang Truces

Little research to date has examined the effectiveness of gang truces and much of the research

that does exist has been restricted to the field of public health Studies of a gang truce in South

Central Los Angeles between the Crips and Bloods by Cotton (1992) and Ordog et al (1993

1995) found temporary reductions in the number of homicides and gunshot wounds during the

truce but these studies failed to point out that despite about a 35 percent decrease in

homicides for the first three months of the truce

homicides then doubled in months four through

eleven compared with the pre-truce period

Similar findings were reported in Trinidad and

Tobago (Maguire Katz and Wilson 2013) where

violence declined for a brief period of time

(again for about three months) but then

increased substantially over the long term (12

months) These results suggest that gang truces

may produce short-term benefits yet result in

long-term adverse consequences

Additionally these studies have suffered from a

variety of flaws including 1) poor theoretical

assumptions about gangs including often

ignoring that they might have the organizational

structure and internal discipline to enforce their

truces 2) relatively weak evaluation designs that fail to account for other potential causes for

What is a Gang Truce

A gang truce is a nonviolent resolution to a larger conflict between groups that has an impact on general levels of violence and other forms of criminality within a community (Ordog et al 1993 1995 Whitehill et al 2012) It differs from conflict interruption resolution or mediation efforts which seek to rapidly intervene typically through outreach workers or violence interruption specialists in episodic violent events between groups in a community Gang truces often involve dialogue and negotiations between multiple parties (eg gangs government NGOrsquos religious organizations) that seek to recalibrate the norms of conflict within and between groups for the purpose of reducing or eliminating violence and other crime

iii

or broader trends related to violence reductions 3) a failure to examine the processes involved

in creating gang truces 4) a failure to examine unintended consequences of the truces beyond

their impact on gun violence and 5) a lack of sophisticated statistical analysis capable of

discerning overall trends cyclic patterns outliers and turning points

Overview of Study Design

This policy brief examines gang truces implemented in El Salvador Jamaica and Honduras

including the processes undertaken with and between gangs and other stakeholders We

collected information about the processes associated with each gang truce from a variety of

sources including peer reviewed articles books reports local newspaper articles and

interviews of persons with first-hand knowledge about the gang truce in each nation We

examined the impact of each truce using official data We first performed a simple t-test

comparing the homicide rates before and after the truce We then employed time series

models not used in other studies to evaluate homicide rates as a function of time with truce

period indicators included to measure the effect of the truce net of the temporal trends We

also employed supplemental models to examine and control for factors other than the truce

that might have affected homicides over the study period

Research Findings

Implementing a Gang Truce

The gang truces studied have a number of common characteristics First in each case a

community was experiencing an uncharacteristically high number of gang related homicides

over a fairly lengthy period of time which resulted in each community placing strong pressure

on the government in general and the justice system in particular to respond to the problem

quickly and effectively Second each community had first attempted unsuccessfully to control

gang violence through suppression-oriented strategies Third each communityrsquos inability to

exercise traditional informal and formal social control to decrease levels of violence became

self-evident to the public and government This resulted in all involved stakeholders wanting to

seek (or participate in) an alternative strategy in which brokers would formally andor

informally work directly with gang leaders to establish a truce that would reduce homicides

iv

Key stakeholders involved in the negotiation and establishment of each gang truce were also

somewhat similar In each case examined the leaders of the largest and most violently

involved gangs were willing to participate In each of these cases it was clear that the gangs

sought to collaborate with the brokers not only for the purpose of reducing violence but

perhaps more importantly as a means to gain greater more positive recognition in the

community and to reap some form of benefit to themselves their members and perhaps their

community In each case while not always formally involved government officials were at a

minimum made aware of negotiations and in some cases solicited the assistance of third

parties to broker an agreement between stakeholders In each case it was at least implicitly

understood that the government would ldquolistenrdquo to the gang leaderrsquos expectations and offers

We found that when the government was no longer willing to ldquolistenrdquo to or collaborate with

brokers the truce processes ended abruptly Brokers were typically comprised of a very small

group (ie 2-3) of individuals who were perceived to be ldquohonest brokersrdquo In El Salvador and

Honduras this included a high ranking Catholic Church official a leader from an international

diplomatic organization (ie OAS) and other neutral parties In Jamaica this included a quasi-

governmental organization that had been established for the purpose of brokering negotiations

between gangs for the purpose of reducing violence as well as the local university which had

access to staff who were perceived to be neutral but had an interest in reducing violence near

the university

The strategies used to execute each gang truce were generally similar but had important

differences Each involved a team of brokers working to identify common goals to be achieved

and tangibles that could be delivered to the gang in exchange for the gang achieving the stated

goals They differed by how each partyrsquos promise was delivered In Honduras and Jamaica gang

leaders traded violence reduction pledges for long-term government and social changes such

as the development of substantial public works programs to reduce unemployment In

Honduras and Jamaica the government was asked to develop and deploy large scale social

programming in a short time framemdashsomething for which neither country had a strong track

record In El Salvador by contrast brokers secured promises for immediate changes in gang

behavior in exchange for feasible immediate deliverables from the government For example in

v

exchange for a reduction in gang violence the government agreed to immediately relocate

imprisoned gang leaders to less restrictive prisons and provide them some privileges Following

the successful execution of the first part of the Salvadoran truce which resulted in near term

success for both parties they began to negotiate broader issues that would take longer for both

sides to deliver Our findings suggest that some promised deliverables need to be easily and

quickly delivered by both parties early in the process to achieve trust and serve as a first test

of gang leadersrsquo ability to deliver Stakeholders have only a brief period of time to provide

promised benefits before trust is lost meaning that tangible benefits need to be delivered in

weeks or months not years

The Impact of a Gang Truce

El Salvador We found that El

Salvadorrsquos gang truce had a definite

impact on the homicide rate The

mean number of monthly homicides

declined from about 354 prior to the

truce to about 218 following the

truce for a net decrease of about

136 homicides per month Our data

show that between March 2012 and

June 2014 the truce had saved

about 5501 lives (see exhibit 1)

From a hypothetical stand it is

possible to make the assumption that a number of these deaths averted could have been

transformed in disappearances and therefore they were not counted within the official

homicide statistics However the results from the analysis point out that the number of

disappearances was not significantly related with the change in the global behavior of

homicides Additionally over the period of analysis (January 2010 to June 2014) there was no

significant change between the pre-truce and post-truce periods in the number of thefts

extortions robberies rapes and auto theftsrobberies

Exhibit 1 Forecast of homicides without gang truce

vi

We also found that the gang truce did not result in a homogenous decline in violence across

municipalities About 61 percent of municipalities experienced a decline in homicides but the

decline in violence varied substantially between municipalities We studied this issue further by

examining the impact of the initiative ldquoFree Violence Municipalitiesrdquo and it was found that the

behavior of violence in those municipalities was not significantly linked to the initiative but

rather to the general dynamic of the truce process nationwide Additionally we parsed out the

relative influence of the number of MS13 and 18th Street gang members on the street and in

prison from each municipality Our analyses indicated that following the truce the number of

MS13 and 18th Street gang members on the street in a municipality was not significantly

related to a decline in homicide but the number of imprisoned MS13 and 18th Street gang

members from the municipality was In particular the number of imprisoned MS13 gang

members from a municipality was associated with a significant decline in homicides in that

municipality following the gang truce and the number of imprisoned 18th Street members

from a municipality was associated with a significant increase in homicides in that

municipality following the truce These findings lend support to the idea that MS13 is more

organized than the typical street gang and that imprisoned MS13 gang members exhibit strong

influence over their fellow gang members on the street Our findings also suggest however

that the gang truce had a boomerang effect in municipalities with high numbers of imprisoned

18th Street members implying that 18th Street might not have as much organizational capacity

to regulate violence on the streets as MS13 The truce provided incarcerated MS13 and 18th

Street gang leaders an opportunity to negotiate with high-ranking officials and influential

diplomats including representatives of the Organization of American States This may have

increased their legitimacy inside and outside of their gangs It appears that MS13 was able to

exert its span of control over the communities in which they had influence and they were able

to deliver on the terms of the gang truce negotiations In the case of 18th Street however

incarcerated gang members may not have had the same organizational capacity for

communicating and carrying out directives In fact a review of the gang truce indicated that

there was a conflict taking place between two factions within 18th Street Consequently the

organizational structure and culture of 18th Street might be more diffuse than that of MS13

vii

and its leadership structure might not be as strong because of the internal fractures within the

gang This might further explain why homicides increased in 18th Street communities The

internal fractures within the 18th Street gang may have resulted in intra-gang violent conflict

that was largely contained within 18th Street controlled territories

Jamaica In Jamaica at first glance our impact findings appeared to show that the gang truce

might be an effective mechanism for reducing violence Bivariate analyses showed a significant

decline in homicides immediately after the truce was implemented This explains the work

previously published by policymakers researchers and news reporters Upon closer

examination of the data however comparing change in the target area to other areas in

Jamaica and accounting for temporal trends we found that the decline in homicide was part of

a larger nationwide decline in violence and that the gang truce was not responsible for the

decline The only significant effect that we uncovered was the possibility that homicides were

displaced outside the target area for a brief period of time but then returned to previous

levels

Honduras Our impact findings from our analysis of data from Honduras tell a very similar story

as Jamaica Initial analysis showed that the number of homicides on average declined across

municipalities following the gang truce Specifically the mean number of homicides declined by

12 per 100000 population from an average of 687 per 100000 population in each

municipality before the truce to an average of 566 thereafter However after we examined the

effect of the truce through time series analysis and included a variable (month) to control for

the temporal trends in the data the impact of the truce we observed in our bivariate analysis

was no longer significant Our findings as in Jamaica suggest that the decline in homicides was

less a consequence of the gang truce than of a broader short to medium term trend And

importantly in both Jamaica and Honduras the respective governments failed to deliver on

gangsrsquo demands for large-scale social and employment programs

Conclusions and Policy Implications

Over the last several years there have been a number of naturally occurring experiments

involving gang truces in a variety of nations and regions Findings from evaluations of gang

viii

truces are mixed In El Salvador the gang truce could be characterized as an effective short

term strategy to reducing homicides It is worth mentioning that despite homicides rates are

above truce levels they continue slightly below pre-truce levels In Jamaica and Honduras the

gang truce had no short or medium-term impact on overall violence In Los Angeles and

Trinidad y Tobago there was evidence that violence decreased for at least ninety days but then

increased substantially beyond those rates observed prior to the gang truce Thus the

effectiveness of the truce in El Salvador appears to be isolated and must be evaluated within

the context of other truces that have failed to reduce homicide violence Policy makers must

evaluate whether the conditions that allowed short term effectiveness of the gang truce in El

Salvador (such as the ability to promise and deliver immediate results) exist in other violent

areas before evaluating whether a truce strategy might be appropriate And they should be

heavily cautioned that the potential for long term negative consequences might outweigh the

potential for short term benefits

Indeed it is important to note that a number of scholars have noted that gang truces are likely

to result in a boomerang effect with gang violence increasing over the long run because of

enhanced cohesion within the gang (Klein 1995) Maguire (2013) notes that when government

officials negotiate a truce with gangs they might ldquoinadvertently be acknowledging gangs as

legitimate social entitiesrdquo (p 11) This in itself might increase cohesion among gangs which has

been found to be associated with increased levels of criminality (Decker et al 2008 Klein 1971

Maguire 2013) Hence it is important to consider the fact that gangs are illegal groups in El

Salvador and it should be cautious when carrying out dialog or negotiation processes with

them Further research is needed to examine how gang truces might impact group cohesion

and if it does whether the cohesion created could be effectively directed toward more

productive non-violent endeavors Gang truces convey the well-intentioned image that violence

has been addressed and policymakers are doing something about the problem but unless the

truce is implemented in a manner and under conditions where immediately achievable results

can be promised delivered and measured there remains a significant chance that the truce will

fail or worse yet backfire Thus it is imperative that any type of concession made by

ix

Governments to gangs within a truce framework should be transparent so that all sectors of

society have certainty that every action is being done within the existing rule of law2

In the case of El Salvador the truce arises from to the absence of effective public policies and

practices for violence control and prevention The truce was planned as a strategy to reduce

gang-related homicides During the process different organizations got involved including

religious (facilitating and protecting human rights) non-government (managing and facilitating

dialog and negotiation processes) international (providing funding for insertion programs) and

government (facilitating and providing certain conditions for dialog and negotiation)

organizations Some of the concessions that the Government provided in order to achieve a

reduction of homicides were within the law but others generated confusion and they seemed

to be close to the legal or socially acceptable limits This fact along with the poor transparency

of authorities in the management of the practice with public media and public weakened the

process and postponed its continuity

The present study suggests that gang truces should only be used as a means of last resort

and then only under certain conditions Given the risks associated with a gang truce

communities with high levels or at least modest levels of formal social control should rely on

other more promising gang control strategies such as pulling levers (ie Boston Ceasefire)

community oriented policing and the Gang Resistance Education and Training (aka GREAT)

program Only when the state has limited or greatly reduced capacity for social control should a

truce be considered Concomitantly a gang truce should be considered as an alternative only

when a community is experiencing a substantial amount of gang violence Communities that

are experiencing minimal to modest amount of gang violence may risk more from the

establishment of a gang truce than they have to gain Additionally our findings suggest that a

gang truce might only be feasible when gangs are sufficiently well organized to be able to

regulate their membersrsquo behavior and cause their members to behave less violently In El

22 Pragmatically this issue is complicated On the one hand transparency is a foundational element within a democracy and is necessary to ensure proper oversight of the government On the other hand it might not be possible to implement a gang truce with too many actors having a voice Policymakers might consider creating a policy that allows such negotiations take place but requires particular actors (such as a judicial body) to be informed of the process to ensure transparency and adherence to the rule of law

x

Salvador there is evidence of the strong organizational structure of gang MS13 Among other

factors the magnitude of its membership the chain of command from its leaders in prison and

the discipline of its leaders in the streets seemed confirmed On the contrary the

organizational structure of gang Barrio 18 ndash divided in two factions fighting over the leadership

ndash showed to be a less stable counterpart within the truce

Finally dialog and negotiations processes with or between gangs must have the capacity to

promise and deliver immediate benefits to the gangs that gang members can see or experience

in order to secure their continued participation in the truce as well as the capacity to monitor

and respond to truce violations Most importantly any effort aimed at reducing violence is

important and should be examined and assessed but it must have a transparent foundation

especially when it affects population rights as a whole

General recommendations

Gang truces are conjunctural strategies States who suffer from gang-related violence must

establish permanent public policies for crime control and prevention A government that

considers implementing a gang truce should be aware that it cannot become the center

strategy of its public policy for citizen safety

Gang truces should only be used as a means of last resort and then only under certain

conditions Stakeholders must determine whether a process of dialog or negotiation with gangs

is legal ethical and feasible

Stakeholders must anticipate demands that are likely to arise and their response options Some

demands may be easily met such as improved prison conditions Others are much more

difficult and amorphous such as community development through more integrated violence

prevention programs (such as those implemented by SolucionES in El Salvador) local economic

development programs or economic reinsertion of ex-gang members

Stakeholders should incorporate immediately achievable and demonstrable deliverables Long-

term goals and promises are unlikely to create the trust needed to sustain a gang truce

Stakeholders must first determine the position in which they are negotiating the incentives that

are possible to deliver and the boundaries and limits they face Gangs are mostly likely to trust

xi

representatives from NGOs community-based organizations and members of the faith based

community as brokers because they are considered more reliably neutral advocates for peace

They need to understand the capacity of the government to deliver promises in a timely

manner

Governments have to make a choice about the visibility and transparency of its participation

This decision needs to be made in the context of the national and local laws the publicrsquos

expectations of transparency and patterns of practices of the past

Governments must be strategic in their support for truce initiatives Some donor funded

programs run by the government prohibit gang member participation and if the government

does not receive approval from the donor it may risk the donor withdrawing its sponsorship of

the program

Governments must ensure an inter-institutional coordination for the management of truces to

avoid the responsibility to be of a single government institution It is necessary to generate or

collect reliable and pertinent data that can be used to analyze and assess the process

It is necessary to implement an effective monitoring system of the truce process similar to that

used in the full report as it can help parties understand what is working and who is delivering

on their promises More specifically the monitoring and truce management system should be

able to identify truce violations and be prepared to respond through the use of legal and

effective practices if stakeholders do not comply

Finally it is necessary to develop evaluations of gang truces and monitoring programs and

support violence prevention activities local economic development activities and pilot

programs to support the reinsertion of ex-gang members into society Clearly national

governments municipal governments NGOs and community-based organizations need

increased capacity and resources to discourage the growth of gangs among at-risk youth It is

therefore increasingly important to create economic opportunities for gang members willing to

leave the gangs and find other legal employment Developing and sustaining those

opportunities in nations with high incidences of poverty will require significant international

funding

xii

References

Cotton Paul 1992 ldquoViolence Decreases with Gang Trucerdquo JAMA 268(4) 443-44

Decker Scott H 2003 Policing Gangs and Youth Violence Belmont CA Wadsworth

Decker Scott H and W Reed 2002 Responding to Gangs Evaluation and Research

Washington DC National Institute of Justice

Decker Scott H Charles M Katz and Vincent J Webb 2008 ldquoUnderstanding the Black Box of

Gang Organization Implications for Involvement in Violent Crime Drug Sales and

Violent Victimizationrdquo Crime and Delinquency 54 153-72

Esbensen Finn-Aage and D W Osgood 1997 ldquoNational Evaluation of GREATrdquo US

Department of Justice Office of Justice Programs National Institute of Justice

Katz Charles M and Webb Vincent J 2006 Policing Gangs in America New York Cambridge

University Press

Klein Malcolm W 1995 The American Street Gang New York Oxford University Press

mdash 1971 Street Gangs and Street Workers Englewood Cliffs NJ Prentice-Hall

Maguire Edward Charles Katz and David Wilson 2013 ldquoThe Effects of a Gang Truce on Gang

Violencerdquo Unpublished paper Washington DC American University

Maguire Edward 2013 Research Theory and Speculation on Gang Truces Woodrow Wilson

International Center for Scholars in Washington DC

McCorkle R and T Miethe 2002 Panic The Social Construction of the Street Gang Problem

Upper Saddle River NJ Prentice-Hall

Ordog Gary J W Shoemaker Jonathan Wasserberger and Michael Bishop 1995 ldquoGunshot

Wounds Seen at a County Hospital Before and After a Riot and Gang Truce Part Twordquo

Journal of Trauma-Injury Infection amp Critical Care 38(3) 417-19

Ordog Gary J Jonathan Wasserberger Julius Ibanez Michael Bishop Eduardo Velayos

Subramaniam Balasubramanium and William Shoemaker 1993 ldquoIncidence of Gunshot

Wounds at a County Hospital Following the Los Angeles Riot and a Gang Trucerdquo Journal

of Trauma 34 779-82

Whitehill J M Webster D W amp Vernick J S 2012 ldquoStreet conflict mediation to prevent

youth violence Conflict characteristics and outcomesrdquo Injury prevention injuryprev-

2012

1

Introduction

Given the increasingly devastating individual family and community effects of gang activity

over the past several decades an increasing body of literature has focused on gangs gang

members and gang activity A core theme running throughout this body of literature is that

gang members are significantly more likely to be the offenders and victims of violent crime than

non-gang members (Battin Hill Abbott Catalano and Hawkins 1998 Esbensen Winfree He

and Taylor 2001 Katz et al 2011 Huff 1998 Curry et al 2002 Miller and Brunson 2000

Miller and Decker 2001 Pyrooz et al 2012 Curry et al 2001) and disproportionately affect

neighborhood levels of crime and violence (Block 2000) These findings have been robust in

that they have been repeatedly found regardless of research methodology (ie use of official

data self-report data observational data) (eg Curry 2000 Decker 1996 Deschenes and

Esbensen 1999 Esbensen et al 2001 Gordon et al 2004 Katz Webb Schaefer 2000 Webb

Katz Decker 2006) or research setting (ie North America Europe Asia South America

Australia) (Klein and Maxson 2006) As a consequence it should not be surprising that

policymakers academics and community leaders have focused much of their attention on

developing responses to address community gang problems

Suppression strategies have been the favored public policy response to gangs since the 1980s

(Venkatesh 1999 Spergel et al 1995) Suppression strategies typically rely on focusing

criminal justice resources on gang members through such practices as targeted and enhanced

police patrols intelligence databases vertical prosecution and enhanced sentences for those

convicted (Katz and Webb 2006) Suppression strategies are based on deterrence theory and

are founded on the principal that swift certain and severe penalties for those involved in gangs

and gang activities will necessarily result in fewer individuals joining gangs and will deter people

from engaging in gang crime (Klein 1995) By the early-to-mid 1990s as gang problems

continued to proliferate and policymakers sought alternative gang control strategies gang

prevention programming flourished (Papachristos 2013) Gang prevention programs were

aimed at the general youth population or focused on at-risk youth or neighborhoods Gang

prevention programming was based on the premise that by reducing risk factors and increasing

2

protective factors prevention specialists could inoculate youth from gang membership

(Esbensen 2000) Prevention programming has most frequently come in the form of school-

based prevention programming such as the Gang Resistance Education and Training (GREAT)

program where students are exposed to a curriculum designed to reduce both participation in

gangs and gang crime as well as the SolucionES3 program which engages youth and their

families through integrated community-based and school based programs

While a burgeoning body of literature has developed describing the assumptions issues and

effectiveness of suppression (eg Decker 2003 Decker and Reed 2002 Katz and Webb 2006

McCorkle and Miethe 2002) and prevention strategies (Esbensen and Osgood 1997) much less

attention has been given to examining gang intervention programming This might largely be

the consequence of the absence of such programming over the past 30 to 40 years By the

1980s policymakers no longer believed that social intervention approaches were an effective

strategy to control gangs and gang problems Although gang intervention strategies took many

forms they were based on the assumption that gang membership is the by-product of a socially

deprived community and that the values and norms of gang youth can be influenced and

directed toward those of mainstream society As such intervention programs often focused on

diverting youth from gangs or sought to minimize the consequences of gangs and gang activity

Gang intervention strategies include crisis intervention dispute resolution street-level

counseling and youth outreach (Spergel 1995) However many policymakers community

organizations and academics have argued that such approaches not only did not reduce gang

activity but may have lead to increased group cohesiveness which in turn may have lead to

increased delinquency (Klein 1971 Spergel 1995) More recent research has yielded similar

results For example a number of studies examining replications of Chicago CeaseFireCure

3 SolucionES is a Global Development Alliance in which $20 million is provided by USAID and SolucionES Alliance

members will contribute an additional $22 million dollars during a 66-month period starting July 2012 SolucionES is being implemented by an Alliance of five leading Salvadoran non-profit organizations who have come together to prevent crime and violence The Alliance members are Fundacioacuten Nacional para el Desarrollo (FUNDE) Fundacioacuten Salvadorentildea para la Salud y el Desarrollo Humano (FUSAL) Fundacioacuten Crisaacutelida (known locally as Glasswing) Fundacioacuten Salvadorentildea para el Desarrollo Econoacutemico y Social (FUSADES) and Fundacioacuten Empresarial para el Desarrollo Educativo (FEPADE) Together these five organizations have notable expertise in education health community development economic development research and youth leadership they are bringing their combined synergy and strengths to prevent crime and violence in El Salvador More information on social violence prevention programming is presented in Annex E including a summary of the violence prevention initiatives being implemented by SolucionES

3

Violence which relies heavily on crisis intervention dispute resolution street-level counseling

and youth outreach have found these strategies to either be ineffective or worse increase

levels of violence (Fox et al 2014)

One type of gang intervention gang truces however has received little attention in the

literature This is somewhat surprising given its use and public claims of effectiveness The

purpose of this report is to systematically examine the effectiveness of gang truces Gang truces

have largely been a ldquoblack boxrdquo that is little is known about the frequency of their use

conditions that give rise to them the role of third parties in brokering gang truces the

transformative effects of truces and the effectiveness of gang truces In this report we review

prior research on gang truces and present evidence on the effectiveness of gang truces that

have been implemented in El Salvador Jamaica and Honduras

Theoretical and Policy Rationales for Gang Truces

Much of the concern about gangs over the past twenty years has been their close association

with violence Qualitative and quantitative research has repeatedly found that gangs and gang

members are involved in high levels of serious violent offending Decker (1996) attributes the

gang-violence relationship in part to the collective and normative structure of gangs which

supports and encourages the use of violence both preemptively and in retaliation He further

explains that gang membership encourages the use of violence in retaliation against threats

and attacks whether actual or perceived which results in patterns of inter-gang conflict

characterized by escalating violence Related Cooney (1998) points out that gangs are not all

that different from ldquowarrior societiesrdquo He argues that while gangs are violent the violence

that they engage in is different than that engaged in by non-gang members in that it commonly

takes on a feud-like dimension A perceived slight violation of turf or other disrespectful

action might invoke a shooting which in turn evokes a retaliatory shooting which in turn

results in another retaliatory shooting creating a self-perpetuating cycle of violence

Early research and theoretical work examining gangs and gang conflict suggested that much

violent gang behavior was the function of status management (Thrasher 1927 Whyte 1943

Short and Strodtbeck 1965) Borrowing from the sociology and social psychology literature on

4

impression management theory gang scholars hypothesized that youth place significant

emphasis on image managementmdashseeking to impress their peers and limiting the potential to

embarrass themselves in front of others (Hughes 2005) These early gang theorists postulated

that violence is an instrument used by gangs and gang members to achieve manage and

protect status They contend that gang members often make decisions to become involved in a

violent conflict based on rational processes that weigh the immediate loss or gain of status

within the gang against the relatively small probability of being formally sanctioned by officials

within the criminal justice system (Hughes 2005) A number of researchers have examined the

relationship between status considerations and gang violence and have found the association

to be particularly robust regardless of gender (Campbell 1991 Giordano 1978) ethnicity

(Anderson 1990 1998 Vigil 1988) and location (Jankowski 1991 Spergel 1995) Hughes

(2005) notes that the centrality of status consciousness by gang youth may account for the

existence of the facilitative gang effect that gang scholars have repeatedly observed (ie Gangs

increase the amount of delinquency individuals are involved in beyond that of a group of

delinquent peers (Gordon et al 2004 Thornberry et al 2003)

Another micro-social factor associated with gang violence is group cohesiveness While our

understanding of the relationship between gang cohesiveness and violence is underdeveloped

some scholars have reported a strong relationship between the two In particular Klein and his

associates (1971 1967) and Lucore (1975) have reported that increases in gang cohesiveness

lengthens periods of gang membership and increases members participation in gang crime and

violence Cooney points out that there are strong relational ties between gang members that

necessarily result in increased cohesiveness among members For example gangs are

comprised of neighborhood youth who share common cultural and economic experiences and

have often known one another for lengthy periods of time He also points out that gangs are

groups that have strong self-proclaimed and formalized identities (eg colors symbols names

monikers) and have at least some organizational structure Their sense of group is maintained

by their common understanding of their members and friends and their attachment to their

territory (or neighborhood) Decker (1996) notes that the relatively high level of group

5

cohesiveness exhibited by gangs facilitates both collective behavior and liability (For exception

see Hughes 2013)

Accordingly both micro-social factors (ie status management gang cohesiveness) serve to

augment levels of gang violence and make it difficult for third parties to intervene Violence

within the context of gangs serves as a form of informal social control Gangs and gang

members cannot seek assistance from legitimate institutions of social control to solve conflicts

because they would risk losing status (Anderson 1999) Likewise the collective nature of the

gang not only increases potential offenders and targets of violence but also facilitates at the

group level the need for retributive justice on the behalf of injured members Moreover gang

members desire to impress others with their commitment to the group and use violence to

demonstrate their commitment to their group and to increase their status within the group All

of this results in an increased cycle of gang conflict and violence4

Over the past several decades residents and policymakers have expected the police to address

violent gang conflicts However there are several limitations to the police response to gangs

First as noted above many of the same factors that are associated with violent gang conflict

also limit the effectiveness of the police to have an impact on violent gang conflict Gang

members are not going to contact the police to resolve a conflict because it could result in loss

of status and expose them to the police discovering illegal activities that they are involved in

(Katz 2001 Katz and Webb 2006) Related citizens in neighborhoods with gang problems are

also reluctant to call the police out of fear of gang reprisals (Katz 1997 Webb and Katz 2003)

or because they have a poor perception of the police themselves (Katz Choate and Webb

2002) Third the police response to gangs in most communities is a reactive response to a

specific incident after it has occurred rather than a proactive response to intervene in ongoing

disputes between gangs (Katz and Webb 2003 2006) Most police agencies simply do not have

4 There is anecdotal evidence for these hypotheses Hughes and Short (2005) through field observations and interviews with members of 20 Chicago gangs examined the micro-social context of violent-related incidents Specifically they were interested in the specific social conditions that precipitated violence Their analysis revealed that gang disputes associated with retaliation escalated into violence significantly more often (ie 55 of the time) than disputes associated with normative order violations or an identity attacks which only resulted in violence roughly 33 percent of time

6

the intelligence networks required to intervene in gang conflicts until after they have risen to

relatively high levels (Katz 2003) Fourth police suppression strategies have been linked to

increasing gang cohesiveness and possible increases in gang crime (Klein 1995)

Some policymakers and community activists have proffered that an alternative to reliance on

formal mechanisms of social control such as the police is the gang truce (Spergel 1995) The

goal of a gang truce is to reduce or even eliminate violent conflict between those gangs that are

warring with one another As such compared with other strategies that often seek to reduce

general levels of crime a gang truce has the very specific goal of reducing violence between

two or more gangs that are in conflict with one another Unfortunately the literature provides

little guidance on the theoretical assumptions of why a gang truce should reduce inter-gang

violence Henderson and Leng (1999) hypothesize that at the root of gang truces is the notion

that they involve the renegotiation of existing norms within and between gangs The authors

claim that as a violent dispute escalates between gangs leaders and members are placed in the

situation of appearing weak to both members of their own gang and to members of the rival

gang if they do not respond with the appropriate amount of force or if they were to suggest a

peaceful resolution to the dispute Accordingly Henderson and Leng argue that as the cycle of

violence escalates between two or more gangs behavioral norms shift toward the increased

valuation of violence to resolve the conflict because it is the only option readily available to

them A gang truce on the other hand which is often mediated by a third-party is believed to

break the cycle of violence by providing the gangs involved in the dispute with a cooling-off

period (Spergel 1995) In the interim period new norms of expected behavior within and

between gangs are established In other words the cooling-off period is believed to recalibrate

norms of behavior that are more consistent with the security interests of the gang and its

members (Henderson and Leng 1999)

For many of the reasons stated above a number of communities have participated in gang

truces Gang truces have been observed in the United States Central America and the

Caribbean and gang leaders in other nations are considering its implantation (Fahah 2012)

Unfortunately little is known about gang truces We do not know when they come into

consideration how they are implemented whether they decrease increase or even have a

7

significant impact on violence In the below section we systematically review the existing body

of literature on gang truces Appendix A describes the methodology we employed to conduct

the systematic review

Prior Research on Gang Truces

Little research to date has examined the effectiveness of gang truces and of the research that

has been conducted has been restricted to the field of public health The most celebrated

truce to date was Philadelphiarsquos House of UMOJA 1974 gang truce In preparation of the truce

gang leaders in prison were called upon for their support along with key community leaders

from churches businesses schools and the police When UMOJA called for the gang summit it

was reported that 500 gang members and 75 of Philadelphia gangs attended By the

conclusion of the gang summit a 60-day truce was announced which reportedly resulted in no

gang members being killed over the 60-day period However no systematic evaluation of the

gang truce was conducted (Woodson 1981)

The first attempt to evaluate a gang truce to our knowledge was conducted by Cotton (1992)

who examined the results of a gang truce in South Central Los Angeles between the Crips and

Bloods Data provided by the police department indicated that over the 6-week period when

the truce took place drive-by shootings decreased by 48 compared to the same 6-week

period the prior year decreasing from 162 to 85 Likewise gang-related homicides dropped by

62 from 26 to 105

Ordog et al (1993 1995) examined the effects of a gang truce in Los Angeles using emergency

room admissions data Specifically the authors examined changes in the daily and monthly

number of gun shoot wound (GSW) emergency room admissions before during and after a

gang truce The catchment area for the emergency room was 100 square miles The authors

noted that while they were able to clearly identify the date the gang truce began because of

the media coverage that it received there was no specified date that the truce ended and as a

5 The authors did not report whether the drop in gang-related homicides was citywide or in the South Central Los Angeles neighborhood where the gang truce took place

8

consequence after 12 weeks the authors regarded it as called off for evaluation purposes

because it was no longer being discussed in the media and gun shot wounds began to increase

Student t-tests were used to examine changes before during and after the gang truce Ordog

et al (1995) reported that there were approximately 7 GSW admissions per day the 12-months

preceding the truce compared to 45 GSW admissions per day during the gang truce and 126

GSW admissions per day in the 11-months following the gang truce The authors concluded

that their analysis ldquoclearly showed that the institution of a gang truce had reduced the number

of GSW victims seen in an hellip inner city Level I trauma centerrdquo (p 419)6

However it is important to point out that while the gang truce in Los Angeles did decrease

homicides by about 35 percent for the first three months it then doubled in months four

through eleven compared with the pre-truce period - a pattern that the authors did not

discuss Similar findings were reported in Trinidad and Tobago (Maguire Katz and Wilson

2013) where it was determined that homicides declined for a brief period of time (again for

about three months) but then increased substantially over the long term (12 months) These

results suggest that gang truces may produce short-term benefits yet result in long-term

adverse consequences

While the research examining gang truces shows their potential promise and their potential for

greater harm we believe that the findings should be viewed with caution for two reasons

First some of the implicit theoretical assumptions that gang truces are built on may not be

accurate Prior research on gangs suggests that they have limited organizational structure

(Decker Katz and Webb 2008 Decker Bynum Weisel 1998) and have few formal

mechanisms to influence member behavior For example most gangs do not have formal

leaders do not require members to pay dues and members do not make contributions to the

gang for the purpose of developing the gangs infrastructure (ie guns housing etc) A number

of academics have pointed out that ldquosophisticated gang organizations is still largely a product of

the self- or organizational-interested musings of gang leaders certain police officials academic

6 Similar results were presented by Ordog et al 1993 using the same data but analyzing it over a shorter period of time and aggregating the data by month instead of by day

9

researchers media reporters based on very limited hard datardquo (Spergel 1995 79-80)

Therefore even if a truce was successfully negotiated between members of gangs in conflict

much of the academic research suggests that gang leaders do not have enough control over

members to enforce a truce Related prior research on gangs and gang members suggests that

they are primarily comprised of young people with few ldquodiplomaticrdquo skills (Henderson and Leng

1999) The very same organizational and normative features of gangs that result in gang

violence (ie often bastions of young minority male street youth) are those same features that

most likely limit the likelihood of a gang having the capacity to abide by a gang truce In sum

implicit in the theoretical assumptions of a gang truce is that gangs have the organizational and

cultural capacity to create and maintain (at least for a short while) a truce which is inconsistent

with much prior research on the organizational characteristics of gangs

The second reason we believe that prior evaluations of gang truces should be viewed with

skepticism is that they have relied on relatively weak evaluation designs For example prior

studies examining gang truces have not incorporated the use of comparison areas or control

groups The causes of reductions in gang violence found in previous evaluations might be

many For example the Los Angeles riots took place just before the gang truce that Ordog et al

evaluated Zinzun (1997) reported that gang culture and violence changed briefly but abruptly

following the riots because gangs and gang members in part redirected their anger and focus

toward the police As such the decline in GSW admissions may have been the consequence of

an overall city-wide decline in gang violence in the wake of the LA riots Related prior research

has relied on fairly broadly defined outcome measures such as GSW admissions and general

levels of drive-by shootings Such outcome measures lack specificity in terms of attributing

gang violence to the specific gangs involved in the gang truce Evaluation designs used in

previous studies have also lacked specificity in terms of the catchment area where violent

activity was labeled as gang related For example Ordog et alrsquos (1993 1995) outcome measure

included all emergency room admissions for GSW However the emergency room received

patients from a 100 mile square area surrounding the hospital an area that was most likely

much larger than the gangsrsquo territories involved in the truce

10

Additionally prior research examining gang truces has not examined the processes involved in

the creation of the gang truce No context has been provided in terms of the factors that lead

to the gang truces whether the gangs were pushed or pulled into truces whether outside

parties helped mediate the truces or whether on-going mediation was required to maintain the

truce In other words we still do not know the processes related to the formation of a

successful gang truce Instead prior research and evaluation has treated the gang truce much

like a black box where it is described in very general terms but its details are not revealed

Likewise prior evaluations have not examined the effects of a gang truce beyond its impact on

gun violence A number of prior researchers have argued that mediating such activities serves

to legitimize gang leaders increases gang identity for members and results in greater group

cohesion (Haskell and Yablonsky 1982 Klein 1995) As such we do not know if there are any

unintended consequences resulting from gang truces due to the methodologies chosen for

prior research and evaluation Finally prior research on gang truces has not relied on more

sophisticated statistical analysis that has the capacity to not only discern overall trends but also

cyclic patterns outliers and turning points

THE PRESENT STUDY

The present study seeks to understand the effectiveness of the gang truce when negotiations

occur with or between gangs and with government institutions or civil society organizations

We have briefly reviewed the theoretical and policy rationales for gang truces and the

literature and prior research on this form of gang intervention Next we present three case

studies one from El Salvador another from Jamaica and another from Honduras Each offers

evidence about their impact on violence and lessons learned that are unique to their particular

circumstances In conclusion the case studies are considered together and we present their

collective policy implications

11

Case Study A Gang Trucemdash The Salvadorian Experience

Over the past two years the Salvadoran gang truce brokered by an array of local policymakers

and international development organizations has attracted national and international

attention Violence in El Salvador has been at an all time high with a rate of 692 homicides per

100000 population in 2011 (UNODC 2014) making El Salvador one of the most violent nations

in the world (Parkinson 2014) Since 1992 government and non-government actors have been

responding to the rising tide of gang violence using traditional suppression strategies such as

increased policing legislative changes and more severe prison sentences These traditional

mechanisms of formal social control however were proving ineffective if not counter-

productive (Perez 2003)

In response to the inadequacy of traditional strategies stakeholders altered their course in an

effort to radically reduce gang violence in the nation Members of the Funes administration led

a group of negotiators comprised of the Catholic Church a former congressman and the

Organization of the American States (OAS) to help frame the conditions for a possible truce

between the MS13 and 18th Street gangs (Umana de Leon and Tager 2014) In March 2012 a

truce was reached The goal of the gang truce was to reduce violence specifically gang-involved

homicides Included in the terms in exchange for the gangs acting to reduce homicides certain

incarcerated gang members were to be transferred to lower security prisons to receive

sentence reductions and special visitation privileges and to be permitted more communication

with those outside of the prison for the purpose of conducting crisis interventions to mitigate

the violence (Salanegra 2012)7 The gang leaders also agreed to no longer recruit children into

their gangs reduce violence against women give up a small number of guns and continue to

participate in negotiations (Seelke 2014 11-12)

7 It is important to note that there were other discussions that took place during the mediations that included as a local publication noted ldquoProhibition Act Gang send the army to barracks end the police operations in controlled areas by gangs repeal of the figure of the witness criteriado (with criteria of opportunity or ldquowitness under a plea agreementrdquo) and a number of improvements in the quality of life of the inmates and its familiesrdquo Source httpwwwsalanegraelfaronetes201209cronicas9612

12

The present case study seeks to understand the role of negotiations with or between gangs and

government institutions or civil society organizations and their impact on violence in El

Salvador The specific objectives of this case study were to (1) Identify and document the

negotiation processes with or between gangs This includes but will not be limited to

identifying the actors involved in the negotiations the goal(s) of the negotiations and

strategies employed to carry out negotiations (2) Identify the impact of negotiations with

andor between gangs on violence and other forms of criminality and (3) Present conclusions

about the strengths and weaknesses of negotiations and their potential for positively or

negatively reducing crime in Salvadoran society

The Salvadorian Gang Truce

While the origins of the Salvadorian gang truce are somewhat unclear some accounts suggest

that in the fall 2011 the Minister of Security (then David Munguia Payes) mentioned the idea of

starting a dialogue between the gangs to Raul Mijango (a former congressman) Shortly

thereafter Mijango spoke to Monseigneur Fabio Colindres the military chaplin about the idea

They started to communicate with gang leaders in January 2012 From the beginning the

Minister of Security stated that he would not personally communicate with gang leaders and

noted that the official position of the government was not to negotiate with offenders But he

did allude to the fact that the gang truce was part of a new strategy to address the nationrsquos

gang problem (Archibold 2012) The Salvadorian gang truce was multi-dimensional involving a

varying number of actors communication styles and tactics In the section below we discuss

the parties involved in the truce the negotiation process and the strategies used by the gangs

and government in furtherance of the truce

Parties involved in the truce There were a series of persons groups and other entities who

directly or indirectly participated in the negotiation process whether they were negotiators or

collaborators From the beginning negotiations between the MS13 and 18th Street gangs were

considered confidential and were held in strict confidence between the gangs and the

negotiators Beginning in early 2012 the first two mediators to serve were a former

congressman (Rauacutel Mijango) and a priest (Fabio Colindres) These individuals were able to gain

13

the trust of the gangs and also had open communications with the government and

community

As the process moved forward however church leadership changed its initial position and

withdrew its support of the truce This left the priest to participate as a private individual no

longer serving as a representative of the church The mediators had some early success

however as the mediators changed the lines of communication became less clear and less

reliable Still they played an important role in the process Finding individuals whom the gangs

would trust and who could speak for government and community institutions was challenging

Further as government ministers and presidents changed it became less clear whether or not

the government was participating in the dialogue between various parties much of the

confusion in negotiations was related to understanding the governmentrsquos official role in the

process

In the beginning the negotiating parties included the two mediators and MS13 and 18th Street

the two predominant gangs in El Salvador MS13 is the largest gang in the country with about

250 cliques throughout the nation For the most part its first- and second-generation leaders

were in prison There are municipalities in which only MS13 operates and therefore its

members never clashed with 18th Street adversaries MS13 members are largely involved in

extortion violence and intimidation of the public The MS13 leadership has been shown to

have greater control of its organization than most its members have been more disciplined

and despite its size the gang seems to have sufficient and effective communication

mechanisms between cliques and its leadership (ldquoranflardquo or ldquoruedardquo)

The 18 Street gang is somewhat smaller than MS13 Long before the truce process began

factions existed within 18th street which had become deeply involved in internal conflicts over

leadership structure Those differences had become more pronounced over time leading to the

establishment of two movements known as the ldquoSurentildeosrdquo (Southerners) and the

ldquoRevolucionariosrdquo (Revolutionaries) that participated in the peace process but under different

14

representation Despite their differences however for the negotiation process the two

fractions of 18th street established similar demands between themselves and with MS138

As noted above the governmentrsquos involvement in the truce process was never clearly

articulated and it varied with ministerial changes The Funes Administration specifically the

Ministry of Justice and Public Security (MJPS) had been engaged in the negotiation process

early on In 2013 however with the change in MJPS leadership this also changed Tensions

grew over the role of the government in the truce negotiations According to the mediators the

new Minister of Security wanted to dismantle the process and for this reason he blocked the

mediatorsrsquo entry to the prisons where they had been meeting with gang leaders The change in

government leadership brought changes in strategy As with all negotiations mutual trust was

the cornerstone When the personnel and their agendas changed so did the trust in the peace

process

Late in 2012 as negotiations began to focus on the relationship between the gangs and the

community other organizations were called upon to help facilitate this interaction A technical

committee was created to oversee progress towards the violence reduction goal this was

announced at the end of August 2012 The committee moved forward with support from the

Organization of American States (OAS) It coordinated actions intended to improve the living

conditions of incarcerated persons to reintegrate into society those who had served their time

in prison and to prevent violence and provide assistance to victims of violence The committee

operated as a mechanism for coordinating the negotiation process with technical support from

the United Nations Development Program (UNDP)‒El Salvador Interviews with experts

indicated that Fundacioacuten Humanitaria with support from several organizations like Interpeace

and others had also supported the operations of the mediators and served as an important

point of reference for the gangs

In each territory where agreements were made government representatives from some

municipalities were invited to participate in the process as key implementation stakeholders

Between December 2012 and January 2013 the municipalities of Ilopango Santa Tecla

8 httpsalanegraelfaronetes201110cronicas5645

15

Sonsonate and Quezaltepeque became among the first to become part of what later became

known as the Sacred Municipalities (aka violence free municipalities) By November 2013

several other municipalities had become part of this group La Libertad Apopa San Vicente

Puerto El Triunfo Nueva Concepcioacuten Ciudad Delgado and Zacatecoluca

The gangs and the government offered several incentives to one another The gangs offered a

number of actions to reduce the incidence of crime (ie cease of hostilities zero homicides

stop extortions robberies and recruitment of children and youths into gangs peace at the

schools) and the government offered to guide public investments in social policies prioritizing

participating municipalities

The truce process was the product of a convergence of vested interests from different sectors

Several earlier attempts at negotiations had failed likely because of a lack of perceived

authenticity on the part of the actors All parties involved in this truce process came to the

table with well-defined group interests and concerns All sides were experiencing fatigue from

the long history of violence Years of fighting deaths and violence had led the mostly

incarcerated first generation those who were still living and who continued to serve as gang

leaders to seek new ways of exercising social power and influence in prison and the

community

The negotiation process In El Salvador the negotiation process between gangs formally began

on March 8 2012 by a decision of the Government to favor the transfer of gang leaders

imprisoned out of the countryrsquos only maximum security prison located in the municipality of

Zacatecoluca That decision was made by the Minister of Justice and Public Security as part of

an agreement between a team of mediators who served as advisors to the Ministry and gang

leaders primarily from MS13 and 18th Street The negotiations were referred to as a ldquotruce

between gangsrdquo and were characterized by the mediators as ldquoa peace processrdquo or a ldquocease of

hostilities between gangsrdquo

Publicly the gang truce was known as an agreement between gangs and not between gangs

and the government because of fear of how those outside the negotiations might interpret the

governmentrsquos actions From the beginning therefore the government never fully

16

acknowledged its participation in the peace process (through representatives) with the gangs

and that reticence was a major impediment throughout the process This was largely because

of the difficulty in communication between the multiple parties which was done through

mediators and necessarily resulted in additional logistical complexity throughout the

negotiations Regardless it is important to recognize that the government played a decisive

and central role in the origin facilitation and promotion of the peace process

In terms of processes interviews with gang leaders indicated that they would communicate

their ldquodemandsrdquo to the mediators who in turn would communicate the gang leaderrsquos demands

to the government The same process was used for the government to communicate with the

gang leaders After an agreement was reached the parties would make proclamations that

emphasized the agreed upon terms of the negotiations These often took the form of public

statements made through local newspapers For example there were more than twenty public

statements made by the gangs where they clearly articulated that their intention was to

reduce the harm that their acts of violence were causing themselves and the communities (For

a detailed discussion of the public statements see Appendix B) In these public statements the

gangs acknowledged the need for their groups to change Both the mediators and the gangs

made clear that the gangs expected the government to respect and respond to their demands

given the good-faith actions that the gangs had already taken

The government on the other hand did not make any formal public statements about their

end of the agreements reached through the negotiations From the outside it appeared that the

government did not develop a formal or consistent policy regarding the truce and instead

assumed that they could achieve a reduction in homicides through the negotiation of increased

prison privileges for gang leaders At the close of the Funes administration the peace process

and negotiations initiated within its framework had reached a plateau There they remained

waiting for the new government to step in and take action as President-elect Salvador Saacutenchez

had announced his support for the truce during the electoral process Sanchez however

withdrew that support when he assumed the presidency in June 2014 Indeed under the

recently elected Sanchez administration the government changed course and became less

17

willing to engage the gangs believing that the government should not negotiate with criminal

organizations

Indeed it should be noted that responses resulting in increased gang influence have been a

concern over the past several decades--from governmental neglect of the problem in the

1990s through the period of the mano dura and super mano dura of the early 1990s to

2000 and to the present gang truce Many have asserted this claim as truth (ie some

responses will increase the influence of gangs) especially in public discourse This study

however does not directly address this point which may or may not be valid9

Strategies employed during negotiations between the gangs and the government Over the

course of the negotiations between the gangs and the gangs and the government at least two

strategies were employed At the beginning of the negotiations strategies were implemented

similar to that of the historic Salvadorian peace accords Each of the parties with the support of

mediators reached agreements and achieved a resolution to their conflict This model

although useful overlooked an important difference between the current violence in El

Salvador and the civil war of the 1980s In a conflict of civilians with the State the legitimacy of

the ldquoadversariesrdquo arises by the need of the state to recover the rule of law It is certainly

possible to reach peace agreements that resolve conflict between gangs but in the current

case with regard to the government there was no legal or policy justification for executing the

truce process On the contrarymdashgovernment negotiations with a criminal group are relatively

rare in modern democratic societies The lack of a legal or policy framework to work from

limited the governmentrsquos ability to be transparent in its response to gangs and may have

undermined its legitimacy with the public

The second strategy employed by the parties was based on reciprocation and cooperation

between actors Early on it was determined by both sides that if one of the parties abandoned

the peace process or did not hold up its end of an agreement the other party would no longer

participate in the negotiation process (Axelrod 1986) In retrospect during the course of the

peace process the gangs implemented agreed-upon terms and positioned themselves as valid

9 We would like to thank one anonymous reviewer for bringing this issue to our attention

18

partners with the government able to negotiate for what is needed by the government in

exchange for what the gang needed10 For example in public statements the gangs insisted

that they had made a goodwill gesture when they declared a unilateral truce and stopped their

involvement in violence The government responded by transferring incarcerated gang leaders

from a maximum-security prison to ordinary criminal prisons11 Next the gangs offered to

disarm the government responded with increased flexibility on control measures at those

prisons This process continued until the demands from both sides grew in a direction that

challenged each sidersquos capacity and willingness to follow through

For instance mediators through the government asked the gangs to end the extortion of

businesses and individuals which are the primary means of subsistence for Salvadorian gangs

The gangs asked that the government reciprocate by eliminating the ldquowitness under a plea

agreementrdquo from the criminal procedural law which is one of the main weapon laws used for

sentencing gang members for complex crimes such as extortion Each of these requests were

more than the parties could carry out Exhibit 1 outlines the different ldquooffersrdquo and ldquodemandsrdquo

made by each party The offers and demands are divided into those that were believed to be

relatively simple and were ldquoofferedrdquo by the party on its own volition and those that were

believed to be more complex and were demanded by the other party

Exhibit 1 Simple demands and concessions made by gangs and the government

Gang concessions Government concessions

Reduce the homicides by more than 50 Transfer the leaders to prisons with lesser security

Hand over 500 firearms to the authorities Allow visits of the children of the gang members

Reduce violent actions at schools Allow night intimate visits

Stop killings at the prisons Allow entry of fast foods

Do not murder custodians police soldiers or their family members

Facilitate the entry into the prison of gang members let out of prison

Give opportunities to some gang members to withdraw

Suspend the searches by the armed forces on persons at the prisons

10 Source httpwwwsalanegraelfaronetes201209cronicas9612 11 This was useful to facilitate the coordination and the communication between the gang members in prisons and in the streets

19

Make it more flexible for new social plans by government to be established in the communities controlled by gangs

Improve the conditions and health assistance at the prisons (eg place tvrsquos to improve mental health of inmates)

Do not burn buses do not kill bus drivers or collecting agents in buses

Allow the Red Cross and journalists to enter into the prisons

The second agenda presented in Exhibit 2 shows the more complex demands that were

not offered by one side or the other but rather were demands placed on the other party These

demands were such that they required a higher level of authority to negotiate in order to

implement the demanded action These demands were considered critical for the peace

process to continue and the delayed responses on both sides stalled the progress of the truce

and led to the parties questioning the legitimacy of the other side all these demands have yet

to be attained The mediators recognized at the end of 2012 that some of the slow progress

was related to the difficulty of making the transition from offers that could be executed

relatively simply to the more complex demands made by each party One of the gang leaders

summarized the issue ldquoWe are not looking for television sets while all of our people continue

living like shit we are not going to try to do everything that is in our power to decrease as much

violence as possible for one television set There are things we are very clear about this [points

to a television set in the room] is a right that the law grantsrdquo

Exhibit 2 Complex demands and concessions made by gangs and the government

Demands by Gangs Demands by Government

Eliminate the figure of witness under a plea agreement from the Criminal Law

Stop homicides and extortions indefinitely

Create international commission to investigate the human rights violation cases of the gang members and their family members on the part of the State (PNC and FA)

Share the information regarding the whereabouts of the brunt of extortion money whether in country or abroad

Institutionalize external and professional surveillance regarding the behavior of the PNC as regards investigations and gang member arrests

Progressive dismantling nation-wide of the clique structures and turfs

Army should stay in its barracks and definitely suspend their participation in public security tasks

Permanently suspend the orders to murder State security and justice agents and their family members (police soldiers judges prosecutors)

20

Clearing of judges prosecutors and police involved in corruption cases against youths in conflict with the law

Permanently suspend the murders extortions and harm caused to public transportation resources and their workers

Maintained a sustained improvement on the conditions of the prisons

Permanently suspend the recruitment of children and youths and hostilities to educational centers

Guarantee working opportunities for the gang members and their family members through specific programs at the municipalities

Share the information about providers of drugs and illegal arms

Television sets were one of the concessions made by the government from the simple list The

gang leader recognized that their intentions in the negotiations extended beyond improving

prison conditions however the government did not have support nor established mechanisms

for carrying out tasks derived from the more demanding list Further as the government

transitioned to new presidential leadership many of the more simple concessions had already

been made and the new government would have to address more complex demands if

negotiations were to continue

The Salvadorian gang truce is remarkable for several reasons First a number of policymakers

and researchers have claimed that the truce saved a large number of lives and was perhaps the

most successful gang truce in the Western Hemisphere Second the Salvadorian gang truce is

somewhat unique in that it might have lasted substantially longer than any other successfully

negotiated gang truce Understanding the temporal impact of the truce is important to

understanding its future potential Third the Salvadorian gang truce is important because a

number of other counties have sought to replicate it For example following the perceived

(and perhaps real) success of the Salvadorian gang truce the nations of Honduras Belize and

Guatemala instituted or tried to institute a similar type of truce Understanding the impact of

the Salvadorian gang truce will further help us understand the potential impact of such a

process on violence in these other nations

Methods

We examined the impact of the gang truce by merging four separate data sets First data from

the 2007 El Salvador Population and Housing Census provided municipal level measures of

21

percent unemployed percent male aged 10-29 percent female headed households ethnic

composition in-migration income percent urban percent households rented and percent

who had a high school education These data were obtained directly from El Salvadors Ministry

of Economy Second municipal level crime data (ie homicide extortion rape theft robbery

and auto theftrobbery) were provided by the National Civil Police (PNC) by month and year for

the period between January 1 2010 and June 30 2014 Disappearance data was also provided

by the police aggregated by year and municipality Third police gang intelligence data was

provided by the National Civil Police (PNC) Specifically we received the number of police

recorded MS 13 and 18th Street gang members by municipality in 2011 the latest year for

which these data were made available Last we acquired 2011 prison gang intelligence data on

the number of imprisoned MS 13 and 18th Street members by municipality from the

Salvadorian National Bureau of Prisons All four data sources were linked together using a

unique municipality identification number and month Collectively they provide an

opportunity to examine the impact of the gang truce in El Salvador and understand several

competing explanations for any changes that might have occurred

The geographic unit of analysis for the present study is the municipality According to the 2007

El Salvador Population and Housing Census there were 14 departments divided into 262

municipalities (the equivalent of a county in the USA) in El Salvador Of the 262 municipalities

ten (38) of the municipalities were eliminated from the analysis because of missing data

Measures

Dependent variables The dependent variable in the present study is the monthly number of

homicides in each El Salvadoran municipality expressed as the number of homicides in

municipality i at time t Homicide data were collected from January 1 2010 through June

30 2014 for a total of 54 months This includes data for a period of 26 months prior to the gang

truce and 28 months following the implementation of the gang truce As shown in Exhibit 3

there were a total of 14148 homicides over the study period with each municipality averaging

371 homicides per month

yit

22

Exhibit 3 Descriptive statistics

n mean sd

Urban Opportunity Factor 252 111 95

male 10-29 years old 252 1960 171

female-headed household 252 3433 499

unemployed 252 1177 686

Racialethnic heterogeneity 252 17 13

of prisoners MS13 252 1579 3960

of prisoners 18th Street 252 1268 4096

of prisoners MS13_spatial weight 252 1579 3960

of prisoners 18th Street _spatial weight 252 1268 4096

MS13 on the street 252 4146 10646

18th Street on the street 252 2513 8622

Violence free municipality 252 04 20

Truce 14148 371 783

Monthly homicide rate per 1000000 14148 51 49

Independent variables Several independent variables were included in our model Our

measure for the intervention is a dummy variable in which the value of 0 is used to represent

the pre-intervention months and 1 represents the post-intervention months We also

included a dummy variable for each community that participated in the violence-free

municipality initiative (sanctuary municipalities) to address issues of external validity (eg

multiple treatment inference) In other words municipalities that participated in the

violence-free initiative might have experienced or participated in something that either

enhanced or detracted from the impact of the larger gang truce The initiative called by some

the second phase of the gang truce took place about 8 to 9 months after the original gang

truce was negotiated in March 2012 In participating sanctuary municipalities gang members

agreed to stop violence and crime in exchange for a reduction in police operations and night

raids12 The municipalities that participated in the initiative included Santa Tecla

Quezaltepeque La Libertad Ilopango Ciudad Delgado Apopa Sonsonate San Vicente

Zacateatecoluca Puerto El Triunfo and Nueva Concepcioacuten

12 See httpwwwcispesorgblogviolence-free-cities-inaugurated-as-second-phase-of-gang-truce

Tt

Si

23

We included several independent variables related to the presence of gangs by gang for each

municipality Specifically we included count variables on the number of MS13 and 18th Street

gang members who were on the streets and in prison by municipality As noted before the

gang truce was made between a relatively small number of imprisoned gang leaders from both

MS13 and both fractions of 18th Street who agreed to stop street-level gang violence on the

condition that they would be transferred to lower security prisons and granted special

privileges

We wanted to understand the impact of the truce in the context of gang presence

Municipalities with high numbers of MS13 and 18th Street members whether they were in

prison or on the streets should have experienced a greater reduction in homicides because of a

greater span of control over these communities Accordingly municipalities with low numbers

of MS13 and 18th Street members should have experienced a lesser reduction in homicides

because of a limited span of control over these communities As Exhibit 3 (above) shows on

average municipalities had about sixteen MS13 and thirteen 18th Street members in prison

and forty-one MS13 and twenty-five 18th Street members on the street We examined whether

gang members in El Salvador were randomly distributed and found that there was spatial

clustering in the number of MS13 and 18th Street members who were in prison In order to

control for the clustering we included the spatial lag (weight) of the MS13 and 18th Street

members in prison

Control variables A series of control variables known through prior research to be related to

violence in communities were added using the census data We began with municipal-level

measures of percent unemployed percent male aged 10-29 percent female-headed

households ethnic composition in-migration income percent urban percent households

rented and percent with a high school education Initial analysis found that five community

variables were highly correlated and loaded on the same factor As seen in Exhibit 4 these

included the percentage of persons who had moved there from another municipality (in-

migration) average income in the municipality percent of the population living in an urban

area percent of houses that are rented and percent of residents with at least a high school

education We labeled this factor Urban Opportunity

24

Exhibit 4 Factor loadings for urban opportunity

Variables Factor

loadings

In Migration Percent of population moved in from another municipality 0638

Income Average monthly income per household (colones) 0886

Urban Percent of population that is urban 0845

Rented Percent of households rented 0761

Education Percent of residents who have at least a high school education 0742

Percent of variance 6071

Eigenvalue 341

Extraction method Principal Axis Factoring

We also calculated a measure of ethnic-heterogeneity from relevant census data Ethnic

heterogeneity which varies from 0 to 1 was calculated by taking one minus the squared

proportions of the population in each ethnic group (White Mestizo other) As with the

percentages of male population that is 10-29 years old female-headed households and

unemployed persons ethnic heterogeneity has consistently been associated with violence in

general (Kubrin and Weitzer 2003) and with gang violence in particular (Katz Maguire and

Roncek 2002 Rosenfeld Bray and Egley 1999)

Analytic Strategy

National-level analysis Our analysis began with a simple t-test of the mean number of

homicides at the national level before and after the truce providing the most basic omnibus

test of an effect We then presented our time series model to estimate the effect net of

seasonality and temporal trends This analysis included a set of simple ARMA (autoregressive

and moving average) models with two lag periods and one period of a moving average Initial

analysis of the number of homicides by month indicated that partial temporal autocorrelations

existed for two lags The first model used only data prior to the truce This model included a

linear time trend variable and dichotomous variables for each month (except January) We used

this model to forecast the expected number of homicides for the truce period This series of

analyses was for illustrative purposes only as the number of data points used in the forecast

25

was too small to provide meaningful confidence intervals The second ARMA model employed

all data from January 2010 through June 2014 and included a dichotomous indicator for the

truce period The effect of this variable was the average change in the number of homicides

net of seasonality (months) and temporal trends

Municipality multilevel models The analysis at the municipal level presented four challenges to

a typical regression model First since our outcome has a highly skewed distribution with

varying levels of exposure (population) we employed a generalized linear model to capture the

correct distribution In this case we employed a negative binomial distribution rather than a

Poisson model due to the over-dispersion in our outcome created by analyzing monthly

municipal datamdashthat is there were several months and municipalities where no or very few

homicides occurred

The second challenge was our need to measure the between-municipality variation of pre-truce

homicide rates and the program effect A fixed-effects model would have been inappropriate

because it would not have allowed us to estimate these variance components Thus we

employed a multilevel random effects generalized linear model

The third challenge was temporal autocorrelation Analysis at the national level indicated

partial autocorrelations in the first and second lags Although a generalized model does not

allow for auto-correlated residuals we addressed this by entering in the model two lags of the

homicide rate

The fourth and final challenge to this estimation was spatial autocorrelation An examination of

the Moranrsquos I and autocorrelation coefficients (exhibit not shown) indicated a low level of

autocorrelation for each month but many months were still statistically significant Thus we

estimated spatial lags of the homicide rate and entered them into the model

Accordingly we analyzed the data using random effects negative binomial models regressing

the rate of homicide on various months on predictors with both temporal and spatial lags For

clarity we present the model using HLM (Bryk et al 1996) notation At the first level (time) we

estimated the log number of homicides using a negative binomial distribution with predictors

26

that included the truce period calendar month linear month trend a

one period lag of the homicide rate a two-period lag of the homicide rate a

spatial lag of the homicide rate and an over-dispersion parameter which has a

Gamma distribution of Thus the final level 1 model is

At the municipality level (level 2) the intercept is a function of the presence of the

Violence Free Municipality program the log number of MS prisoners the log

number of 18th Street prisoners the log of the spatial lag number of MS prisoners

the log of the spatial lag of the number of 18th Street prisoners the log of the

number of MS members the log of the number of 18th Street prisoners the

control variables detailed above and the random effect Thus the level 2 model for the

intercept is

Also at the municipality level (level 2) the effect of truce is a function of SAFE the

log number of MS prisoners the log number of 18th Street prisoners the log of

the spatial lag number of MS prisoners the log of the spatial lag of the number of 18th

Street prisoners the log of the number of MS members the log of the number

of 18th Street prisoners and the random effect Thus the level 2 model for the

truce effect is

TRUCEt mt TIMEt

HRt-1i HRt-2i

HRti vti

Gamma1

aa

aelig

egraveccedilouml

oslashdivide

mti = expb0 i + b1iTRUCEt + lmmt

m=1

11

aring + l12TIMEt

+l13HRt-1i + l14HRt-2i + l15HRti + vti

eacute

euml

ecircecircecirc

ugrave

ucirc

uacuteuacuteuacute

exp vti[ ] ~ Gamma1

aa

aelig

egraveccedilouml

oslashdivide

b0 i

SAFEi PMSi

P18thi

PMSi P18thi

MMSi M18thi

Ci u0 i

b0 i = g 00 +g 01SAFEi +g 02 ln PMSi +1( ) +g 03 ln P18thi +1( ) +g 04 ln PMSi +1( )

+g 05 ln P18thi +1( ) +g 06 ln MMSi +1( ) +g 07 ln M18thi +1( ) + pC

C

aring Ci + u0i

b1i SAFEi

PMSi P18thi

PMSi

P18thi MMSi

M18thi u1i

27

The key parameters of our analysis are in this expression The truce effect is dependent on

the log number of MS prisoners among other characteristics Thus the percent

reduction in homicides can be calculated from this model using the following formula

which we plot below (see Quantitative Findings below)

To answer questions about the variation in the effects of the gang negotiations we predicted

best linear unbiased predictions (BLUP) of the Truce random effect which estimates the

between-community variance of the immediate effect of the truce Since we assumed that the

interventions were not evenly spread throughout communities in this study the mixed model is

appropriate to address this issue We conducted this analysis for homicide using a Stata

generalized linear mixed model using full maximum likelihood and an identity covariance matrix

of random effects

Findings

Exhibits 5 and 6 present our findings related to the number of police recorded homicides in El

Salvador from January 2010 through June 2014 The trends in the data and the results of the t-

test show that in the 26 months prior to the gang truce there were on average about 354

homicides per month compared with about 218 homicides per month in the 28 months

following the gang truce Exhibit 7 shows that the gang truce resulted in 5501 fewer homicides

than otherwise would have occurred

b1i = g 10 +g 11SAFEi +g 12 ln PMSi +1( ) +g 13 ln P18thi +1( ) +g 14 ln PMSi +1( )

+g 15 ln P18thi +1( ) +g 16 ln MMSi +1( ) +g 17 ln M18thi +1( ) + u1i

b1i

ln PMSi +1( )

1- exp g 10 + ln PMSi +1( )g 12( )eacute

eumlugraveucirc( )acute100

u1i

28

Exhibit 6 Number of homicides in El Salvador by month with ARIMA Model Fit

Exhibit 5 Change in homicides from pre-truce to post-truce

Obs Mean SD 95 confidence interval

Pre-truce period 26 35442 4200 33745 37138

Post-truce period 28 21835 6361 19369 24302

Change -13606 10638 16574

=p lt 05

29

Exhibit 7 Forecast of homicides without gang truce

Next as seen in Exhibit 8 we examined the spatial distribution of the change in the homicide

rate after the implementation of the gang truce The analysis showed that of the 252 analyzed

municipalities 243 (93) experienced a decrease in homicides however within these

municipalities there were wide variations in the degree of the decline For example of the 243

municipalities that experienced a decrease the decrease in the homicide rate varied from

about 59 percent of these municipalities experiencing a 1-74 percent decrease to about 9

percent of municipalities experiencing a 75 percent or higher decrease Additionally a modest

number (n=19 7) of municipalities experienced an increase in their homicide rate

30

Exhibit 8 Percent reduction in homicide rate by number of municipalities

Exhibit 9 presents six negative binomial models for the monthly number of homicides The first

model examines the impact of the gang truce and implementation of the Violence Free

Municipality program and our interaction variable that measures the additive effect of both

the gang truce and the Violence Free Municipalities program on the number of homicides We

found that although the gang truce was associated with a significant decline in homicides the

Violence Free Municipality program was related to a significant increase in homicides13

However countrywide the additive effect of implementing the gang truce and the Violence

Free Municipalities program was associated with a significant decline in homicides

13 To be clear our bivariate analyses showed that the violence free municipalities program was related to a significant decline in homicides For example on the one hand those municipalities that did not participate in the violence free municipalities program experienced a decline in their homicide rate from 406 homicides per 100000 population in the pre-truce period to 321 homicides per 100000 in the post-truce period On the other hand those municipalities that participated in the violence free municipalities program experienced a substantially greater decline in their homicide rate from 752 homicides per 100000 population in the pre-truce period to 392 homicides per 100000 in the post-truce period However after other variables are controlled for in our negative binomial models we found that the violence free municipalities program did not have a positive impact on homicides over and above the gang truce itself

31

Exhibit 9 Negative Binomial models for monthly number of homicides

Model 1 Coefficient (se)

Model 2 Coefficient (se)

Model 3 Coefficient (se)

Model 4 Coefficient (se)

Model 5 Coefficient (se)

Model 6 Coefficient (se)

Homicide - 1 month lag 0010 0010 0010 0010 0010 0010

(0002) (0002) (0002) (0002) (0002) (0002)

Homicide - 2 month lag 0012 0013 0013 0013 0013 0013

(0002) (0002) (0002) (0002) (0002) (0002)

Homicide rate spatial lag 0003 0003 0003 0003 0003 0003

(0001) (0001) (0001) (0001) (0001) (0001)

Gang truce implemented -0483 -0541 -0553 0057 -0544 0564

(0066) (0065) (0064) (0278) (0064) (0279)

Violence free municipality 0497 0183 -0003 -0076 -0037 -0094

(0181) (0167) (0152) (0143) (0153) (0145)

Ln( +1) of MS13 prisoners 0066 0176 0062 0171

(0023) (0028) (0043) (0028)

Ln( +1) of 18th St prisoners 0067 0044 0064 -0043

(0021) (0024) (0021) (0024)

MS13 prisoner spatial lag -0004 0054 -0003 0049

(0045) (0055) (0046) (0056)

18th St prisoner spatial lag 0022 0014 0019 0012

(0027) (0032) (0027) (0032)

Gang truce violence free municipality

-0346 -0294 -0309 -0161 -0304 -0059

(0158) (0156) (0158) (0146) (0158) (0145)

Gang truce Ln( +1) of MS13 prisoners

-0185 -0185

32

(0030) (0030)

Gang truce Ln( +1) of 18th St prisoners

0160 0060

(0025) (0025)

Gang truce Ln( +1) of MS13 prisoners spatial lag

-0139 -0138

(0064) (0064)

Gang truce Ln( +1) of 18th St prisoners spatial lag

0023 0023

(0039) (0039)

Ln( +1) of MS13 on street 0023 0018 0021 0016

(0013) (0012) (0013) (0012)

Ln( +1) of 18th St on street 0043 0039 0040 0037

(0016) (0015) (0016) (0015)

Time 0010 0011 0011 0011 0011 0011

(0002) (0002) (0002) (0002) (0002) (0002)

1month 0064 0065 0067 0067 0067 0067

(0055) (0055) (0055) (0055) (0055) (0055)

2month -0020 -0020 -0022 -0023 -0022 -0023

(0056) (0056) (0056) (0056) (0056) (0056)

3month 0088 0088 0188 0088 0088 0088

(0052) (0052) (0052) (0052) (0052) (0052)

4month -0011 -0011 -0011 -0011 -0011 -0011

(0052) (0052) (0052) (0052) (0052) (0052)

5month 0020 0020 0020 0021 0020 0021

(0052) (0052) (0052) (0052) (0052) (0052)

6month 0000 0000 0000 0000 0000 0000

7month 0036 0036 0036 0036 0036 0036

(0052) (0052) (0052) (0052) (0052) (0052)

33

8month 0047 0047 0047 0046 0047 0046

(0051) (0051) (0051) (0051) (0051) (0051)

9month -0055 -0055 -0057 -0058 -0057 -0058

(0053) (0053) (0053) (0053) (0053) (0053)

10month 0040 0040 0041 0041 0041 0041

(0052) (0052) (0052) (0052) (0052) (0052)

11month 0022 0022 0022 0022 0022 0022

(0052) (0052) (0052) (0052) (0052) (0052)

12month -0040 -0040 -0042 -0043 -0042 -0043

(0053) (0053) (0053) (0053) (0053) (0053)

Urban opportunity factor 0209 0042 0033

(0040) (0040) (0039)

male 10-29 0039 0007 0009

(0021) (0019) (0019)

female-headed household -0006 0002 0000

(0007) (0007) (0006)

unemployed 0001 -0002 -0000

(0006) (0005) (0005)

Racialethnic heterogeniety 0058 -0039 -0087

(0263) (0235) (0230)

Intercept -10571

-10544

-11054 -11618 -11015 -11558

(0064) (0060) (0194) (0241) (0212) (0257)

Ln(Alpha) -2287 -2289 -2293 -2294 -2293 -2294

(0113) (0113) (0114) (0114) (0114) (0114)

Truce random effect 0202 0193 0202 0144 0201 0143

(0039) (0038) (0040) (0031) (0040) (0031)

Intercept random effect 0315 0226 0192 0164 0183 0158

(0047) (0035) (0032) (0027) (0031) (0027)

Cov(Truce and Intercept) -0162 -0099 -0122 -0086 -0114 -0090

(0037) (0031) (0031) (0025) (0031) (0026)

34

The second model includes those variables from Model 1 but also includes variables that

controlled for community-level structural factors In this model the gang truce remains

significantly associated with a decline in homicides but the Violence Free Municipality program

and our interaction variable were no longer related to a reduction in homicide Although the

percentage of residents in a municipality who are male aged 10 to 29 female-headed

households percent unemployed and ethnic heterogeneity were unrelated to changes in

homicide the urban opportunity factor was significantly associated with homicide

Models 3 and 4 included our measures of intervention as well as our measures of the number

of MS13 and 18th Street on-the-street and incarcerated gang members at the municipality level

The analysis showed that while the number of MS13 on the street was unrelated to homicides

the number of 18th Street gang members was associated with an increase in homicide These

analyses also indicated that the number of incarcerated MS13 and 18th Street gang members

from a municipality was associated with a significant increase in homicides We further

examined whether this relationship was associated with the gang truce by including two

interaction variables as presented in Model 4 One measured the interaction between the

number of incarcerated MS13 gang members and the gang truce and a second measured the

interaction between the number of incarcerated 18th Street gang members and the gang truce

The results showed that following the gang truce the number of incarcerated MS13 gang

members was related to a significant decrease in homicides and the number of incarcerated

18th Street gang members was related to a significant increase in homicides

Models 5 and 6 in Exhibit 9 present the same two models (3 and 4) as above but include the

community-level structural variables Model 5 once again shows that the number of MS13

members on the street remains unrelated to the change in homicides and the 18th Street

members on the street was associated with a significant increase in homicide However Model

6 shows that following the gang truce the number of incarcerated MS13 gang members

remained significantly associated with a decline in homicides while the number of incarcerated

18th Street gang members remained significantly associated with an increase in homicides

35

Community-level structural factors were found once again to be unrelated to changes in

homicide rates

Exhibit 10 below further illustrates the relationship between reductions in homicides by

municipality and the municipal-level presence of MS13 and 18th Street gang members in

prison The figure shows that in municipalities with no incarcerated MS13 and 18th Street gang

members no change in homicides occurred following the gang truce However when a

municipality had ten MS13 gang members imprisoned on average that municipality

experienced a 55 percent reduction in homicides When a municipality had fifty MS13 gang

members imprisoned those municipalities on average experienced a 71 percent reduction in

homicide following the truce Conversely our analysis showed that the number of 18th Street

gang members in prison (from a municipality) had a significant and positive impact (that is the

number increased) on homicides following the truce For example if a municipality had ten

18th Street gang members imprisoned on average that municipality experienced a 31 percent

increase in homicides14

14 We examined whether there was an interaction effect between gang members on the street following the gangs truce Analysis for both MS13 and 18th street and their interaction with the gang truce showed no effect

36

Exhibit 10 Impact of the gang truce by number of gang members in prison at the municipal level

The Gang Truce and Crime Displacement

We examined two types of displacement crime type displacement and method displacement

Some policymakers have suggested that although homicides may have declined as a

consequence of the gang truce other forms of criminality or categories of reported crime may

have subsequently increased Crime type displacement occurs when offenders who focus on

one type of crime switch to another form of crime in order to avoid detection or to benefit in

some other way (Eck 1993) In El Salvadorrsquos case some critics of the gang truce have suggested

that as homicides decreased other forms of criminality such as extortion might have

increased substantially due to gang membersrsquo increased freedom to conduct activities inside

and outside of prison (Dudley 2013 Parkinson 2014) Method displacement occurs when

offenders change their tactics or methods of conducting crimes as a consequence of an

intervention (Eck 1993) Following the gang truce in El Salvador some analysts argued that gang

37

members might have begun to hide the bodies of homicide victims to avoid detection and to

ldquoprotect the integrity of the countryrsquos gang trucerdquo (Bargent 2013 1)

As noted above we rely on police data that measured monthly numbers of thefts extortions

robberies rapes and auto theftsrobberies by municipality We also received data on the

annual number of disappearances in each municipality by year Exhibit 11 presents the

descriptive statistics for these variables It shows that the rates of theft and robbery did not

change significantly between the pre- and post-true periods and the rate of extortions

significantly declined The t-tests showed that the rates of rape disappearance and

theftrobbery fromof an auto increased significantly

Exhibit 11 Descriptive statistics for measures of displacement

Mean Std Err [95 Conf Interval]

Theft rate

Pre-truce 1238 049 1142 1335

Post-truce 1213 049 1115 1310

Extortions rate

Pre-truce 353 023 308 398

Post-truce 271 021 229 314

Robbery rate

Pre-truce 544 025 495 593

Post-truce 553 025 504 602

Rape rate

Pre-truce 049 003 043 056

Post-truce 064 005 054 073

TheftRobbery Vehicle rate

Pre-truce 132 022 090 175

Post-truce 165 023 120 210

Disappearance rate

Pre-truce 805 068 670 940

Post-truce 1007 064 882 1132

p lt 05

We further examined the data (excepting disappearance data) similarly to the analyses above

in which we used random effects negative binomial models and regressed the number of

38

crimes (ie theft extortion robbery rape and auto theftrobbery) on various months on

predictors with both temporal and spatial lags Our independent and control variables

remained the same as those used in Model 6 (displayed above in Exhibit 9) We found that

over the study period there was no significant change between the pre-truce and post-truce

periods in the number of thefts extortions robberies rapes and auto theftsrobberies (tables

not shown)

As noted above we also received data on the number of disappearances by municipality and

year but because the data were provided by year there were not enough data points to

examine them temporally Therefore we added this covariate (number of disappearances by

municipality and year) to Model 6 in Exhibit 9 The results of the analysis indicated that the

number of disappearances was not significantly related to change in homicides the gang truce

remained associated with a significant reduction in homicides and our interaction variables

(number of incarcerated MS13 amp 18th Street members truce) remained significantly

associated with homicides

Conclusions

The present study sought to examine the impact of the gang truce on violence and other forms

of criminality We examined homicide data by municipality prior to and after the gang truce

Our outcome variables were obtained from the PNC along with several independent and

control variables obtained from the Ministry of Economy and the National Bureau of Prisons

We analyzed this data using a number of time series and random effects negative binomial

models where we regressed the rate of homicide on various months on predictors with both

temporal and spatial lags and controlled for other contextual factors This type of analysis

allowed us to correct for the number of homicides in one month being related to the number of

homicides in the previous month (ie temporal autocorrelation) Additionally it allowed for

the fact that some communities participated in supplemental interventions (ie violence free

municipalities initiative) which might have impacted the outcome in the same way in a given

month These techniques allowed us to isolate the effects of the gang truce as best possible

We also used data from the PNC that measured by municipality the number of thefts

39

extortions robberies rapes auto theftsrobberies and disappearances to examine the

possibility of the trucersquos impact on crime displacement and diffusion of benefits`

Our findings suggest that El Salvadorrsquos gang truce had a significant and dramatic impact on the

homicide rate The mean number of monthly homicides declined from about 354 prior to the

truce to about 218 following the truce for a net decrease of about 136 homicides per month

Our forecast showed that between March 2012 and June 2014 the truce had saved about 5501

lives As will be discussed further below the analysis suggests that key stakeholders have the

capacity to renegotiate existing norms of violence and that at least some gangs have the

capacity to exert substantial informal social control over their members that can result in

reduced violence

The gang truce also lasted substantially longer than previously evaluated truces Ordog et al

(1995) for example reported that the much publicized gang truce in Los Angeles decreased

homicides by about 35 percent for the first three months but then doubled in months four

through eleven compared with the pre-truce period Similar findings were reported in Trinidad

and Tobago (Maguire Katz and Wilson 2013) where it was determined that homicides

declined for a brief period of time (again for about three months) but then increased

substantially over the long term (12 months) These results suggested that gang truces may

produce short-term benefits yet result in long-term adverse consequences

The findings of the present case study suggests that some gang truces might last longer than

previously believed While the number of homicides began to slowly increase about 12 months

following the truce in El Salvador the results clearly showed that almost two years following

the truce homicides still remained below those experienced prior to the truce

We found however that the gang truce did not result in a homogenous decline in violence

across municipalities About 61 percent of municipalities experienced a decline in homicides

but the decline in violence varied substantially between municipalities For example while

about 16 percent of these municipalities experienced a 25 percent reduction in homicides a

number of others (37) experienced a 75 percent or greater reduction Furthermore it is

important to note that about 20 percent of municipalities experienced a modest increase in

40

homicides This suggests that the impact of a gang truce might be variable and could be

dependent on contextual factors We examined the possible influence of these factors by

assessing the impact of social structural factors and the presence of gangs on the municipal-

level impact of the gang truce While we found that social structural factors were unrelated to a

decline in homicides municipal-level gang presence was associated with the decline in violence

as a consequence of the gang truce

We examined this issue further by parsing out the relative influence of the number of MS13

and 18th Street gang members on the street and in prison from each municipality As noted

above we hypothesized that municipalities where gang member presence was high regardless

of their presence on the street or in prison would experience a greater reduction in homicides

because of their increased influence in these areas We found however that the relationship

was not as direct as we would have suspected In particular our analyses indicated that

following the truce the number of MS13 and 18th Street gang members on the street in a

municipality was not significantly related to a decline in homicide but the number of

imprisoned MS13 and 18th Street gang members was associated with a significant change in

homicides following the gang truce

Of special note was our finding of the differential impact of the truce based on gang Although

we found that the number of imprisoned MS13 gang members was associated with a significant

decline in homicides following the gang truce we also found that the number of imprisoned

18th Street members was associated with a significant increase in homicides following the truce

These divergent findings might be associated with each gangrsquos organizational structure and its

capacity to regulate member behavior

Much prior research suggests that imprisoned MS13 gang members have substantial influence

over violence in Salvadoran communities15perhaps even more so than formal mechanisms of

15 Instituto Universitario de Opinioacuten Puacuteblica (Iudop) La situacioacuten de la seguridad y la justicia 2009-2014 entre expectativas de cambio mano dura militar y treguas pandilleras Aguilar Jeannette (Coord) [et al] --1ordf ed --San Salvador El Salv Instituto Universitario de Opinioacuten Puacuteblica (Iudop) 2014- 260 p graacuteficos tablas cuadros y figuras 28 cm (Talleres graacuteficos UCA) httpwwwucaedusviudopwp-contentuploadslibro_la_situaciC3B3n_de_la_seguridadpdf

41

social control such as the police and courts16 Law enforcement officials for years have claimed

that MS13 is one of the most organizationally sophisticated street gangs in the Western

Hemisphere17 The gang has been widely characterized as having a highly vertical organizational

structure and strong control over criminal enterprises in gang-controlled neighborhoods and as

being decidedly capable of enforcing rules through discipline MS13 leadership resides in the

nationrsquos prison system The Ranfla (gang leadership) is comprised of thirteen MS13 gang

members (PNC 2011) who direct coordinate and authorize street crime and other gang activity

from prison Our findings lend support to the idea that MS13 is more organized than the typical

street gang and that imprisoned MS13 gang members exhibit strong influence over their fellow

gang members on the street

Our findings also suggest however that the gang truce had a boomerang effect in

municipalities with high numbers of imprisoned 18th Street members implying that 18th Street

might not have as much organizational capacity to regulate violence on the streets as MS13

The truce provided incarcerated MS13 and 18th Street gang leaders an opportunity to

negotiate with high-ranking officials and influential diplomats This may have increased their

legitimacy inside and outside of their gangs It appears that MS13 was able to exert its span of

control over the communities in which they had influence and they were able to deliver on the

terms of the gang truce negotiations In the case of 18th Street however incarcerated gang

members may not have had the same organizational capacity for communicating and carrying

out directives In fact a review of the gang truce indicated that there was a conflict taking place

between two fractions within 18th Street Consequently the organizational structure and

culture of 18th Street might be more diffuse than that of MS13 and its leadership structure

might not be as strong because of the internal fractures within the gang This might further

explain why homicides increased in 18th Street communities The internal fractures within the

18th Street gang may have resulted in intra-gang violent conflict that was largely contained

16 Instituto Universitario de Opinioacuten Puacuteblica (Iudop) La situacioacuten de la seguridad y la justicia 2009-2014 entre expectativas de cambio mano dura militar y treguas pandilleras Aguilar Jeannette (Coord) [et al] --1ordf ed --San Salvador El Salv Instituto Universitario de Opinioacuten Puacuteblica (Iudop) 2014- 260 p graacuteficos tablas cuadros y figuras 28 cm (Talleres graacuteficos UCA) httpwwwucaedusviudopwp-contentuploadslibro_la_situaciC3B3n_de_la_seguridadpdf 17 httpwwwlaprensagraficacomeua-declara-a-ms-grupo-delictivo-transnacional

42

within 18th Street controlled territories Further analysis is needed to examine this specific

issue

While not the primary focus of this case study we controlled for any impact that the Violence

Free Municipalities program might have had on homicide in 11 municipalities during the truce

As noted above the Violence Free Municipalities program served as the second phase of the

gang truce (Cawley 2013) The program was first proposed by the two mediators and designed

by the Technical Committee on Violence and Crime Reduction which included representatives

from OAS MJPS the mediators and the Humanitarian Foundation (CISPES 2013) The program

involved the mayors of each municipality collaborating with gang leaders to design prevention

and intervention resources for gang members and at-risk youth The Minister of Justice and

Security David Mungia Payes announced that his ministry would facilitate $74 million in

funding from the OAS UN and other donors to implement the programing In exchange gang

leaders agreed to end violence and other criminality in the Violence Free Municipalities (CISPES

2013) The negotiators also agreed to further discuss both gangsrsquo demand to repeal of the

2010 law that increased the capacity of the police and prosecutors to crack down on gangs

(Ayala 2014)

Our findings suggested however that the Violence Free Municipalities program was unrelated

to change in homicides in these communities While much additional research is needed to

understand why the program was not effective it might have been because the gangs had

already agreed to a truce and had already reduced violence to the extent that they could

Conversely the demands made by the gangs as part of the Violence Free Municipalities

program might have been more than could be delivered The time and resources required to

implement the programming and the political capital that was required to repeal legislation

might have been much more than could be delivered Future research is needed to examine the

processes and impact of the Violence Free Municipalities program

Regardless the base rate of violence in El Salvador was reset for a period of time suggesting

that perhaps the gang truce substantially altered existing norms of violence Klein (1995)

argued that cycles of gang violence (that is perceived or real changes in gang activity) are

43

typically the consequence of seasonality epochal variation (peaks and valleys in violence) and

illusion (it appears as if there is a real change in violence but there is not) In El Salvador we

appear to have observed a self-directed cycle of normative change wherein incarcerated MS13

gang leaders directed a reduction in violence by actuating their organizational span of control

through the gang truce Some of the most powerful and influential gang members in El Salvador

used their political social and economic capital to promote the truce and articulate new norms

of violence They were able to do this by leveraging their informal social control over the streets

through actual or perceived threats of violence against those who violated the terms of the

truce

However following a change in government leadership and the government subsequently

distancing itself from the gang truce the conditions and capacity of MS13 leaders to intervene

in local violence might have deteriorated and violence began to increase substantially This is in

part because the third parties in the negotiation were no longer able to communicate with

government officials about furtherance of the truce Specifically the mediators were no longer

able to negotiate on the behalf of the government and were no longer permitted entrance to

the prisons where they could negotiate with gang leaders

44

Case Study B Gang Trucemdash The Jamaican Experience

In the latest report by the UN Office of Drugs and Crime Jamaica ranked as the sixth most

violent nation in the world with a homicide rate of about 521 per 100000 residents (UNDOC

2011) Policymakers and researchers have attributed the nationrsquos high level of violence to such

factors as drug trafficking (Klein Day and Harriott 2004) access to illegal firearms (Lemard and

Hemenway 2006) and historical processes that include a legacy of conflict between the

nationrsquos two primary political parties (Sives 2002 Figueroa and Sives 2003 Moser and Shrader

1999) all of which have facilitated the entrenchment of the more powerful gangs in

communities of the urban poor Likewise high levels of income inequality and chronic youth

unemployment (Francis and Lyare 2006) problematic urbanization (Stone 1975) social

marginalization and an emergent subculture of violence (Harriott 2008) have contributed to the

nationrsquos violence problem Whatever the causes violence has had a considerable impact on

Jamaicarsquos social and economic development it has decreased investor confidence (Schwab and

Porter 2008) tourism (Harriott 2007) and access to public services It has also increased the

costs related to the health care system (Mansinghand and Ramphal 1993) the criminal justice

system (Caribbean Human Development Report 2012) and the education system (Moser and

Holland 1997)

Jamaicarsquos homicide problem is closely associated with its gang problem The Jamaican

Constabulary Force (JCF) has estimated that some 272 gangs are active in the nation most in or

near the capital city of Kingston (Harriott 2014) Gang types and their respective historical

patterns of conflict matter in Jamaica as these variations may determine their predisposition or

amenability to lasting rather than opportunistic truces Jamaican gangs include territorially

organized crime groups conflict gangs defense crews who regard themselves as defenders of

their communities (Levy 2009) and other less cohesive more transient territorial groups Some

of the latter are predatory others bond around the taken identities that generate conflicts of

other similar groups18

18 These are generally rooted in subcultural issues such as the demand to be treated with respect or with the indiscretions and self-centered aggressiveness of individual members that may be associated for example with sexual competition for the favors of women

45

Estimates of the gang-related homicide rate in Jamaica vary perhaps because the crime is not

clearly defined as far as attributing a death to a gang Regardless researchers agree that the

proportion of homicides that are gang related has increased substantially For example Harriott

(2003) reported a fourfold increase in the rate of gang-related homicides between 1983 and

1997 Likewise Hill (2013) using official police data found an eightfold increase from 2001 to

2009 reportedly in 2001 only about 64 percent (n=73) of the nationrsquos 887 homicides were

gang related compared with 523 percent (n=879) of its 1682 homicides in 2009

To address the problem Jamaica has initiated traditional law enforcement strategies such as

establishing a specialized gang unit (Sinclair 2004) initiating curfews (Sinclair and Tuner 2005)

declaring states of emergency (Jamaican Observer 2010) and implementing community-

oriented policing (Kolpack 2006) It also attempted legislative reforms to curb election fraud

and electoral-related violence that involved local gangs (Levy 2009) None of these strategies

stemmed the tide of gang violence In 2002 the Minister of National Security established the

Peace Management Initiative (PMI) (Henry 2011) to augment governmental and non-

governmental organizational capacity to settle gang disputes in the community through

intervention-based programming such as ceasefires and gang truces

The current study examines the peace initiative instituted by the PMI in Greater August Town

Our objective was to understand the negotiation processes undertaken with and between

gangs and other stakeholders Among other things we were interested in identifying the actors

involved in the negotiations the goal(s) of the negotiations and the strategies employed to

carry them out Most importantly however we wanted to determine whether the gang truce

resulted in the desired outcome a reduction in the number of homicides in the Greater August

Town area

The Greater August Town (Jamaica) Peace Initiative

Greater August Town is located on the northeastern outskirts of the city of Kingston The low-

income area has high rates of youth unemployment and a history of gang-related violence

(Charles 2004 Levy 2009) Over the last decade the Greater August Town community has

sought improved living conditions and revitalization (Levy 2009 95) The arearsquos inherent

46

resilience has been augmented by nearby intellectual and cultural engines such as the

University of West Indies the University of Technology and University Hospital (Charles 2004

38)

Greater August Town is comprised of the communities of August Town proper (which is

fractured into several locales with gang-given names such as ldquoVietnamrdquo and ldquoOpen Landrdquo and

city government-approved names such as Hermitage Goldsmith Villa (Angola) and Bedward

Gardens These socially defined community divisions and subdivisions are markers for the

territorial boundaries of street gangs and therefore in some instances are lines of potential

conflict Some of those boundaries demarcate areas of Greater August Town that are

predominantly supportive of one or another political party but the boundaries do not always

hold political significance politics is but one element in the conflict geography of the area Like

many communities of the urban poor the Greater August Town area is easily mobilized

politically ndash a reality that is understood and at times exploited by street gangs who politicize

gang ldquowarsrdquo in their efforts to build alliances and to neutralize the police In fact the basic

principle of community mobilization in Jamaica is political patronage and clientelism Access to

resources (eg jobs housing education) for the poorest residents is often determined by the

local political party Thus according to Charles (2004 36)

supporters attach themselves to the political parties to get first preference in the

distribution of scarce resources and over time because they are unemployable they

become dependent on their political party for their economic survival These supporters

will kill anyone who threatens the support base of their political party because they

perceive it as a threat to their daily survival

As a consequence some political supporters invest heavily in the electoral contests and

provoke conflicts that affirm their loyalty to their party in order to secure material benefits

from it Political competition is one conflict fault line in what otherwise is a politically

heterogeneous community Specifically political support in Greater August Town is divided

between the Peoplesrsquo National Party (PNP) and the Jamaica Labor Party (JLP) The PNP receives

strong support from residents in August Town proper the upper region of Goldsmith Villa and

47

Bedward Gardens while the JLP is supported by those living in Hermitage and the lower region

of Goldsmith Villa (Charles 2004 ) The division between the political parties in the area as in

much of Jamaica was believed to be at the root of much of the communityrsquos violence

particularly between 1979 and 1993 Some gangs aligned with the PNP and others with the JLP

this often resulted in political boundaries overlapping with gang turf (Charles 2004)

Pre-truce Violence

Nationally the history of gang violence in the Greater August Town area first appeared as

political violence closely associated with the electoral cycle As in many other urban

communities the problem intensified mdash deeply affecting community life in the period just prior

to the national elections of 1980 mdash then continued cyclically until 1993 Gangs have since

harnessed this legacy in order to establish and maintain community support based on common

political affiliations They have used that support to nullify the efforts of law enforcement to

suppress their illicit activities (Harriott 2008)

The conflict profile of these gangs and of the communities in which they are nested has

changed over the decades From the beginning of their involvement in political violence and

territorial control a form of gerrymandering existed that manipulated the voting population

forcing certain individuals out of a particular constituency and preventing those who remained

in the community from voting for the opposing party This was linked both to the electoral cycle

and to the types of inter-gang conflicts that typically were associated with street gangs

(Figuerou Harriott and Satchell) More recently however much of the crime and violence

perpetrated by gangs has resulted from internal conflicts (eg status management disputes

over womengirls or money individual membersrsquo activities that could attract police pressure)

Internal conflict at times has led to gang fragmentation and new alliances that pull more parties

into the conflict escalating the homicide rate and increasing the sense of insecurity among the

general population (Levy 2012)

The most significant of these internal conflicts resulted from the killing of former Jungle 12

leader Neil Wright by members of his own gang Jungle 12 was the most influential gang in

Greater August Town Before his murder in order to increase the gangrsquos access to illicit

48

opportunities in Kingston Wright had been attempting to extend Jungle 12rsquos influence with a

system of alliances with other gangs and recruitment in Kingston (Harriott 2014) In short his

ambition was to transform Jungle 12 from a neighborhood street gang operating on the

outskirts of Kingston into to a dominant organized crime network that could reach into the

heart of the city In pursuit of this goal Wright recruited members from outside August Town

elevating them in the gang hierarchy above the locals This led to status-related conflicts and

resistance to Wrightrsquos leadership within the gang His murder precipitated a split of Jungle 12

into three factions two of them fled to other neighborhoods within August Town (Angola and

Vietnam) resulting in the formation of new alliances and a new conflict geography that

replaced the former political geography of conflict Wrightrsquos killing and the subsequent demise

of Jungle 12 as the dominant gang in Greater August Town altered the balance of power and

escalated inter-gang violence (Harriott 2014) The post-2005 phase of conflict was

characterized by power symmetry conflict intensification and the spread of conflict

throughout the entire geographic area of Greater August Town

Although their origins are unclear retaliatory killings and other violent incidents progressively

intensified between 2005 and 2008 The violence was episodic retaliations were most often

motivated by suspicions related to personal and geographic connections between warring

gangs As the violence escalated new alliances were formed to enhance power and dominance

which in turn increased the number of gangs and gang members involved in the violence

(Harriott 2014) This eventually attracted national attention and triggered community

mobilization for a gang truce

The Truce-making Process

The Greater August Town gang truce was preceded by frequent intense violence and public

outrage As noted above the violence had escalated in November 2005 when Jungle 12 leader

Neil Wright was killed The defection of a Jungle 12 member to Goldsmith Villa (Angola) caused

infighting within the gang and conflict between it and Goldsmith Villa Just a few months later

Wrightrsquos brother Steve and two others were injured during a turf battle (Martin-Wilkens 2006)

49

Thereafter violence began to occur at regular intervals until January 2007 when the Peace

Management Initiative (PMI) hosted a peace march in the community Two PNP politicians

urged the community to unite A PMI leader declared that the peace march was being held to

ldquodemonstrate to the public that Jungle 12 members are back together and that they want

peacerdquo (Thompson 2007 1)

Although hopeful some residents remained skeptical about the peace march perceiving the

action to be politically motivated In the absence of trustworthy information inter-group

conflicts tended to be interpreted through a politically partisan lens this created obstacles to

isolating the gangs building a consensus for peace and unifying community mobilization The

politically based narratives weakened the communityrsquos leverage for peace as well as the

exposure of the gangs to police action As one resident said ldquoThe election is coming up and

they want[ed] the people to vote for the PNP is one of the main reasons why they have to walk

todayrdquo (Thompson 2007 1) Those who shared such views stayed away from the peace march

Although that widely held myth was not factual it did serve to demoralize and demobilize one

part of the community A local UWI faculty member was articulate in his assessment of the

politics behind the march

August Town violence is not violence of organized crime which is based on drug

trafficking extortion or some other criminal enterprise [but] rather the violence in

August Town is essentially ldquotribalrdquo mdash the Peoplersquos National Party tribe versus the

Jamaica Labour Party tribe [which has been complicated by a splintering within the

PNP tribe] (2007 1)

Indeed it was not the violence of organized crime but neither was it political violence The

individual quoted above neglected to mention that the conflict was between those pro-PNP

splinter groups who were largely comprised of members of Jungle 12 Moreover their ldquopro-

PNP-nessrdquo was unrelated to the conflict there was no factional infighting within the local PNP

organization at that time

Nevertheless following the peace march the gang violence diminished Then in November

2007 a turf war erupted between two gangs from the Greater August Town neighborhoods of

50

Vietnam and River This time as the police stated the gang violence was less about politics and

more about dominance and turf Police were dispatched to perform directed patrols but

whenever they were not present the shootings continued (Mcleod 2007) In April 2008 the

community witnessed local gangs engaging in a five-hour-long street battle that left two killed

and three others wounded It ended only after the police deployed armored trucks The next

month another round of gang violence resulted in five others being killed including a one-year-

old child This resulted in the three members associated with the gang who committed the

homicides being killed in retaliation (Virtue 2008) The local community mobilized against the

violence increasingly cooperating with the police providing more information about the gangs

Subsequently gang members observed a decline in their influence within the community

During the early period characterized by low-intensity conflict the less influential gangs at

times used manipulation of the police as a tactic for suppressing the more influential gangs

This was largely done through strategic release of information Prior to 2005 when Jungle 12

was dominant its membersrsquo illicit activities were constantly reported to police by members of

other gangs as a means of compelling a compromise or settlement of conflicts In practice this

was done by ldquotrading casesrdquo Once a crime had been investigated by the police and suspects

had been charged an opportunity was created for the gangs and other parties to the conflict to

settle the matter by agreeing to drop their cases (typically by ceasing cooperation with police

investigators) This type of ldquoself-helprdquo served to end some of the retaliations but it rested upon

the manipulation of the police (Harriott 2014)

Later in an attempt to quell escalating inter-gang violence the police established buffer zones

between the warring gangs This action resulted in unintended consequences For example

when the police declared a buffer zone between August Town and Hermitage Hermitage took

advantage of the opportunity to attack Angola Some Angola residents accused the police of

turning a blind eye and creating an opportunity for Hermitage to attack their community

Although little reliable information exists about why the police made the deployment the way

that they did it is more likely that the police inadequately assessed the situation (ie mis-

assessed the pattern of alliances and the likely targets of attack) (Harriott 2014) In the areas

affected by this kind of increasing violence community members became angered and lost

51

confidence in the police The error resulted in some parties to the conflict receiving increased

support from their communities and in greater gang-community cohesion (Harriott 2014)

After a brief period the police identified this problem and began to disengage by no longer

providing a buffer between gang controlled areas which in turn allowed still more conflict to

occur between the gangs

As the violence escalated beyond their control police finally responded by applying their own

forms of pressure For example units under the direction of the JCF High Command would

make periodic raids in the community during which they would at times seize weapons and

make mass arrests (Sinclair 2005) However there were also moments when the local police

were very responsive improved their relations with the community and consequently gained

greater access to relevant information Two such moments occurred just prior to and again

immediately after the truce moments during which there was greater freedom of movement

and open collaboration between the community and the police (Harriott 2014)19

The Establishment of the Greater August Town Gang Truce

The Greater August Town gang truce was led by the Jamaican Peace Management Initiative

(PMI) The PMI is a government-funded initiative created for the purpose of working with gang

members to reduce violence Due to community mistrust of the police in 2002 the organization

was established as an alternative organizational mechanism for responding to gang violence

The PMI sought to bridge government and civil society efforts to mediate disputes between

gangs as well as to provide outreach to gang members (Bakrania 2013) While efforts to

institute a gang truce in Greater August Town were led by the PMI a number of other

stakeholders helped to facilitate the truce these included faculty at the University of the West

Indies (UWI) and representatives from the police the faith-based community and the August

Town Sports and Community Development Foundation (Jackson 2008 Levy 2009 also see

Appendix C) The gangs involved in the truce included those from August Town Hermitage

19 The quality of police-community relations largely depended on the style of the local station commander however regardless of the external environment

52

Goldsmith Villa Bedward Gardens and African Gardens Because of its formality as well as its

perceived effectiveness the truce signed on June 24 2008 was regarded by many as the first

of its kind in Jamaica (Levy 2009)

Truce negotiations began early in June 2008 and lasted for about three weeks The gangs

sought to leverage their violence-making capabilities and demanded payment for peace They

asked the third-party negotiators for money ldquoworkrdquo and start-up funds for proposed micro-

businesses (Wilson 2014) Those demands were rejected by the negotiators on the grounds that

the third-party institutions would not buy a peace that was intended to save the lives of those

who were making the demands Moreover if peace was to be purchased then gang conflict

could be used continuously to extract money and other benefits from negotiators The third-

party actors made some demands of their own In some quarters of the community and society

the surrender of guns was viewed as a litmus test of the sincerity of the gangs Consistent with

this the negotiators suggested that the parties to the conflict symbolically hand over one gun

each that suggestion was immediately rejected by the gang leaders These kinds of demands

from the various parties ceased after a time as they all agreed that the truce was to stand on

its own merits (Harriott 2014)

As the truce began to be committed to paper a number of stakeholders expressed concern that

their greatest risk in participating could be the potential for Jungle 12 factions to use the peace

agreement as they had in the past as a tactic to persuade their enemies to let their guard

down Others however recognized that Jungle 12 had now been weakened and that a formal

public peace agreement would be beneficial to the gang and therefore this time would be

different (Harriott 2014)

The gangs held fast to their claim that their weapons were needed for their own protection

because the police were ineffective in responding to violence in their communities (Jamaican

Gleaner 2014) It became a precondition of the truce that the gangs would not be required to

turn in their guns and other weapons (Jackson 2008) The truce agreement did specify

however that ldquoall persons are allowed to move freely across all boundaries regardless of

reputation or affiliation No gun salute or any other shooting is to take place in the community

53

for a period of at least five yearsrdquo (2008 also see Appendix C) The truce agreement and its

conditions were prescribed in a document that was finally signed by all of the major

stakeholders including the gangs (see Appendix C)

Throughout the negotiations each of the gang leaders had attempted numerous times to use

the truce as an opportunity to bargain for money jobs and business support grants Such

demands consistently were rejected by the third-party actors Nonetheless both prior to and

after the truce some efforts were made to create better opportunities for young people

residing in the community UWI for example provided a homework supervision program to

encourage students to further their education and it developed a community-building initiative

to help improve schools and enhance sports programming (Levy 2009) Such programs were

conducted as part of UWIrsquos Township Project in August Town which invested significant

resources in developing the residentsrsquo job-related capabilities and collective self-efficacy

The Greater August Town gang truce was noteworthy for two reasons First the gang truce

received substantial press attention The media were invited to witness the ldquosigningrdquo of the

truce by the gang leaders in the presence of a JCF Deputy Commissioner of Police a PMI board

member and the UWI Principal and two professors of its faculty Second the truce was widely

credited with decreasing violence in Greater August Town and it served as an exemplar to

other communities seeking to replicate its success (Virtue 2008) A number of reports

manuscripts and newspaper articles proclaimed the truce to be a success Bakrania (2013 10)

for example reported that ldquoPMI has been credited with stopping gang wars in August Town

rdquo Levy (2009 94) remarked that the ldquomost interesting outcome of PMI efforts to date was the

Peace Agreement reached in August Town in late 2008rdquo Likewise a government report noted

that ldquothe peace treaty was a pivotal achievement in August Town that has significant potential

for wider application Crime levels dropped markedly in August Town after the signing of the

peace agreement in June 2008rdquo (McLean and Blake-Lobban 2009 78) To this day August Town

celebrates the signing of the truce with an annual celebration with food and music

(Cunningham 2011)

54

Methods

Our evaluation relied on a pre-testpost-test quasi-experimental group design Our

methodology examines the Greater August Town community which is comprised of three

contiguous towns where the gang truce took place (the target area) and the balance of Jamaica

which is comprised of 178 communities (comparison areas) As seen in Exhibit 12 the average

number of residents living in each of the three communities in the target area was not

significantly different than that for the rest of Jamaica about 7776 residents lived in each of

the Greater August Town communities compared with 6468 in the other communities

Likewise communities of Greater August Town were about as densely populated as other

communities (2960 per square kilometer versus 2647 per square kilometer) and the age

range of residents was similar as well However Greater August Town (a) had a significantly

higher proportion of its residents living in poverty (196 vs 158) (b) consumed fewer

resources than other communities and (c) reported significantly more homicides than other

communities prior to the truce (see Exhibit 12)

55

Exhibit 12 Descriptive characteristics of Greater August Town and balance of Jamaica (2007-2011)

Comparison

Area Greater August

Town All areas

Population (mean) 6468 777633 648994

(sd) 720482 353731 715621

Population density (mean) 264719 296033 265238

(sd) 271023 285501 270465

Percent in poverty 1577 1957 1583

(sd) 1036 106 1029

Consumption 15737890 1106939 1566048

(sd) 10713020 205336 1064021

residents under 15 yrs old 2369 2494 2371

(sd) 487 115 484

residents 15-65 yrs old 6840 6901 6841

(sd) 423 29 419

Murder per month (mean) 674 857 677

sd 1928 1409 1920

Total murders 10068 180 10248

n 178 3 181

plt=05

Measures

Two distinct data sets were merged to measure the impact of the Greater August Town truce

First data from the 2011 decennial census provided community-level measures of the social

and economic characteristics of the 181 communities in Jamaica Described in detail below the

community-level data used in the study included population population density gender age

poverty and consumption20 These data were obtained directly from the Statistical Institute of

Jamaica

Second police homicide data from the years 2007 through 2011 were used to construct the

studyrsquos community-level measure of homicide The homicide data were aggregated by month

20 Consumption is an alternative measure of poverty in Jamaica which measures the consumption of food and non-food items

56

and appended to the community-level data The final (merged) data set included 10248

homicides over the 60-month study period These data were obtained from the Jamaica

Constabulary Force (JCF)

The dependent variable examined in the study was constructed from official police homicide

data Once again the homicide data represented the number of officially recognized homicides

in Greater August Town and each of the remaining communities in Jamaica We examined

change by comparing the homicide data 18 months prior to the truce with the homicide data 42

months following the truce More specifically we examined whether there was a change in the

number of homicides in the 30 days following the truce (month 1) as well as whether the truce

had an impact every three months thereafter (ie months 2-5 6-8 9-11 12-14 15-42) and

whether any changes in homicide coincided with changes in homicide in the balance of

observation areas The frequency distribution of our dependent variable is presented in Exhibit

13 It shows that prior to the truce the target area on average experienced significantly more

homicides (1495) than did the comparison areas (920)

57

Exhibit 13 Distribution of homicides in the target and comparison areas

Comparison

Area Target

Area Total

Pre-truce period Mean 920 1495 932

SD 2469 1966 2461

N 241400 5100 246500

Month 1 of truce Mean 741 286 733

SD 1785 496 1772

N 17800 300 18100

Months 2 thru 5 of truce Mean 647 905 652

SD 1821 1249 1812

N 71200 1200 72400

Months 6 thru 8 of truce Mean 577 1236 588

SD 1690 1074 1683

N 53400 900 54300

Months 9 through 11 of truce Mean 718 333 711

SD 2034 733 2019

N 53400 900 54300

Months 12 through 14 of the truce Mean 687 095 678

SD 1519 286 1509

N 53400 900 54300

Months 15 thru 42 of the truce Mean 564 589 564

SD 1683 1042 1674

N 516200 8700 524900

Total Mean 674 857 677

SD 1928 1409 1920

N 1006800 18000 1024800

An illustration of the trends in homicide prior to and following the gang truce are shown in

Exhibit 14 It shows that 30 days following the truce homicides fell in the target and

comparison areas then increased and decreased several times with a general downward slope

in violence over time

58

Exhibit 14 Monthly number of homicides pre-post truce in the target and comparison areas

We also used a number of measures to control for community-level structure from the 2011

decennial census These community-level data included the communityrsquos population

population density (per square kilometer) and community level of consumption Additionally

the census data included measures of the percentage of the population that was female under

15 years old 15 and 65 years old and 65 years old and older as well as a measure of the

percentage of the population living in poverty Principal components analysis was used to

reduce some of these data into a summary measure

Exhibit 15 shows the results of the component loadings One component was extracted that we

designated as socio-economic status (SES) which exhibited high loadings for percent living in

poverty percent under 15 years old percent 15 to 65 years old and consumption Excluded

from the principal components analysis were population and population density Population

was used as our exposure variable and population density was logged to address skewness in

these data

Exhibit 15 Factor loadings from principal components factor analysis

Loading Poverty 78 Consumption -76 under 15 years old 92 between 15 and 65 years old -80

59

Analytic Strategy

In order to test whether the truce had an impact on homicides in the target area andor

whether displacement had occurred in the balance of the study area several analytic

techniques were employed Most of the methods employed the use of the homicide rate as the

dependent variable We explored the data in this way to provide the maximum statistical

power to detect an effect As a check on these methods we also employed a generalized model

to compensate for the non-normality of our outcomes

First focusing only in the target area we performed a simple t-test comparing the homicide

rates before and after the truce (the unit of analysis was a month) however this technique had

limitations the most severe of which was that even if the test were significant it would be

difficult to determine whether the difference was due to the gang truce or a natural change

over time in the outcome Second to address this limitation time series models were

employed whereby the homicide rate for the target area was modeled as a function of time

with truce period indicators included to measure the effect of the truce net of the temporal

trends These models were estimated with ARIMA techniques with a one-month lag auto-

correlated error Third we examined the homicide rate for each town using a panel time series

model In this model the temporal trend for each town was examined with indicators for target

areas and truce periods included The main effects for the truce periods measured the effect of

the truce in the target areas and the moderators of the truce period in the comparison areas

measured displacement effects Finally because the dependent variable coded is not normally

distributed across months we used a negative binomial time series model to estimate the

number of homicides with the population covariate serving as an exposure variable

Findings

The first set of results examines only the target area The first test was a simple t-test

comparing the mean homicide rates before and after the truce periods The result was a mean

difference in the homicide rate of -890 per 100000 with a significant t-statistic of 370 While

60

this result is statistically significant we caution that it may or may not reflect an impact of the

truce To further examine the truce effect in the target area we performed ARIMA regressions

The first model did not include an effect of the temporal trend In this model we again find that

by month 15 the murder rate decreased by about -89 per 100000 (Exhibit 16)

Exhibit 16 Results of basic ARIMA model

Next we employed the ARIMA model again but included a variable (date) to control for the

temporal trends in the data Exhibit 17 shows that when we controlled for temporal trends the

impact of the truce we observed was no longer significant This result indicates that it was not

the truce per se that caused the decline in homicides but instead the decline in homicides was

part of a larger (local and nationwide) decline in homicides

Note The test of the variance against zero is one sided and the two-sided sigma 7877339 5907266 1334 0000 6719536 9035142 L1 1720758 1242218 139 0166 -0713944 415546 ar ARMA _cons 1538436 1904285 808 0000 1165203 1911669 t_15 -8888017 2969383 -299 0003 -147079 -3068134 t_12 -1507749 1820406 -083 0408 -507568 2060181 t_9 -132475 9189969 -144 0149 -3125951 4764507 t_6 -2861008 6335492 -045 0652 -1527835 9556328 t_2 -5166363 4498142 -115 0251 -1398256 3649834 t_1 -1111771 4463623 -025 0803 -9860311 7636769murders_rate murders_rate Coef Std Err z Pgt|z| [95 Conf Interval] OPG

Log likelihood = -2089969 Prob gt chi2 = 00397 Wald chi2(7) = 1472Sample 564 - 623 Number of obs = 60

ARIMA regression

61

Exhibit 17 Results of ARIMA model with control of temporal trends

We next estimated the possible displacement effects of the truce Exhibit 18 presents the

results of these models Examination of the main effects of the truce period does not indicate

any effects and looking at the truceComparison interaction effects we also do not find any

displacement effects Note that these models also controlled for the socio-demographic

characteristics of each community

Note The test of the variance against zero is one sided and the two-sided sigma 7775398 6622814 1174 0000 647735 9073446 L1 1270618 1253989 101 0311 -1187156 3728392 ar ARMA _cons 1480274 1197324 124 0216 -8664378 3826986 date -2318963 2091772 -111 0268 -6418761 1780835 t_15 -3519687 7694691 -005 0964 -1543329 1472935 t_12 -970747 2228007 -044 0663 -5337561 3396067 t_9 -8844212 9402792 -094 0347 -2727335 9584921 t_6 2995697 6576644 005 0964 -1259042 1318956 t_2 -2353533 4923798 -048 0633 -12004 7296933 t_1 -9194625 5462116 -017 0866 -1162501 9786088murders_rate murders_rate Coef Std Err z Pgt|z| [95 Conf Interval] OPG

Log likelihood = -2082031 Prob gt chi2 = 00710 Wald chi2(8) = 1444Sample 564 - 623 Number of obs = 60

ARIMA regression

62

Exhibit 18 Results of Panel (Town) time series model with control of temporal trends

Last we used a random effects negative binomial regression that predicted the homicide rate

with population as an exposure variable and controlled for the socio-demographic

characteristics of each community The results are presented in Exhibit 19 The analysis showed

that time had a negative effect indicating that homicides were decreasing in all areas over the

study period The main effects of the truce (truce = 1 2 ) represented the effects of the

truce in the targeted area and did not show a significant effect for any period following the

gang truce However we did find that the homicide rate significantly increased in the

_cons 6310151 3518738 179 0073 -5864483 1320675lnpopulationdensity 121961 1709831 713 0000 8844891 1554731 ses 1245334 2428629 513 0000 7693312 1721336 date -1276623 0345659 -369 0000 -1954101 -0599144 15Balance 5392222 4128863 131 0192 -2700202 1348464 12Balance 1318516 7903159 167 0095 -2304746 2867507 9Balance 1042658 7925876 132 0188 -5107854 2596101 6Balance -1703928 7919006 -022 0830 -1722489 1381704 2Balance 3498997 7175155 049 0626 -1056405 1756204 1Balance 1056874 1104696 096 0339 -1108291 3222039 truceaugust Balance -4757021 3276618 -145 0147 -1117907 1665032 august 15 -4269255 4284086 -100 0319 -1266591 41274 12 -1262496 7871907 -160 0109 -2805361 2803697 9 -1014276 7881965 -129 0198 -2559113 5305603 6 6563019 786731 008 0934 -1476334 1607595 2 -5034368 7123478 -071 0480 -1899613 8927392 1 -1215158 1095699 -111 0267 -3362688 9323715 truce murders_rate Coef Std Err z Pgt|z| [95 Conf Interval]

Prob gt chi2 = 00000 Wald chi2(16) = 12359 max = 60 avg = 5661878Estimated coefficients = 17 Obs per group min = 48Estimated autocorrelations = 1 Number of groups = 181Estimated covariances = 1 Number of obs = 10248

Correlation common AR(1) coefficient for all panels (02045)Panels homoskedasticCoefficients generalized least squares

Cross-sectional time-series FGLS regression

63

comparison areas in months 12 through 14 following the truce In particular we found a 29

percent increase in the homicide rate in the comparison communities for that period (exp (-

1797 + 2048) = 1285 p-value = 004) Since this effect is only significant at the 005 level

however and given the number of analyses used to examine the data it is possible that we

found this effect by chance alone

64

Exhibit 19 Random Effects Negative Binomial

65

Given these caveats we visualized this model with the following set of marginal predictions as

observed in Exhibit 20 We saw that the targeted area (as illustrated in red) experienced an

immediate decrease in homicide which coincided with an increase in homicides in the balance

of the study area However the target area quickly returned to ldquonormalrdquo and homicides in the

comparison area decreased again During months 9 through 11 following the truce there was a

reduction in homicides in the target area with an associated increase in the comparison area It

is important to point out that the confidence intervals are large and we cannot yield concrete

conclusions from these results However it appears that the truce might have had a temporary

short-lived displacement effect decreasing homicides in the target area but increasing

homicides in the comparison area

Exhibit 20 Predicted change in homicides in the target and comparison areas

Conclusions

From 2000 through 2009 Jamaica experienced a substantial number of homicides many of

which were attributed to gangs in one form or another Traditional law enforcement responses

were repeatedly implemented but until 2010 those had little effect Some policymakers in

Jamaica as well as in other nations throughout the Caribbean and Central America have

recently been experimenting with novel approaches to reducing gang-related violence notably

the implementation of gang truces In Jamaica at least eight gang truces reportedly have been

66

negotiated since 2001 (Levy 2009) The Greater August Town gang truce was thought to have

been one of the more successful and it has served as a model for other communities to use

(2009) Our purpose here has been to identify the actors involved in the negotiations of that

truce the negotiation goals and the implementation methods used and then to examine

empirically the impact of that truce on homicide rates in the targeted community

The 2008 gang truce in August Town was a response to violence that arose when the leader of

one gang was killed creating a power vacuum that other gangs saw as an opportunity to

increase their influence in the community Concomitantly the community as well as the gangs

feared that an absence of formal social control would result in further violence The police

reacted unevenly At some times they engaged in appropriate but heightened levels of

preventive patrol while at other times they purposely provided little or no protection on

occasion they used aggressive tactics that further isolated them from the community The end

result was that there was neither stability nor predictability in the police response and

therefore little trust in the police to address the problem

As the violence further escalated the community mobilized The Jamaican Peace Management

Initiative faculty members from the University of the West Indies (UWI) the Jamaican

Constabulary Force (JFC) and other community-oriented groups joined forces seeking to

reduce the increasing number of homicides by brokering a truce between the gangs Over the

three-week negotiation period the negotiators and the gangs sought terms from one another

The gangs wanted payment ldquoworkrdquo and funds for micro-business development to end the

violence The third-party stakeholders wanted the gangs to disarm actually or symbolically

Neither the gangs nor the stakeholders had substantial leverage nor did they have much to

offer one another in terms of incentives In the end however a truce was agreed upon and all

of the gangsrsquo leaders and several key community stakeholders signed it at a public ceremony

with the media in attendance

At first glance our impact findings appeared to show that the gang truce was an effective

mechanism for reducing violence Bivariate analyses showed a significant decline in homicides

after the truce was implemented This explained the work previously published by

67

policymakers researchers and news reporters Upon further examination of the data however

comparing change in the target and comparison areas and accounting for temporal trends we

found that the decline in homicide was part of a larger nationwide decline in violence and that

the gang truce was not responsible for the decline The only significant effect that we

uncovered was that possibly the homicides were displaced outside the target area for a brief

period of time but then returned to normal

Any one of a number of explanations might be offered for the strategyrsquos lack of effectiveness It

might be that the Jamaican gang leaders at least those in Greater August Town did not have

the organizational capacity to change gang member behavior Much prior research suggests

that in general gangs have limited organizational structure and little formal leadership This

might suggest that gangs do not possess the necessary capacity to regulate their membersrsquo

violence That said gangs in Jamaica including in Greater August Town have been found to be

fairly organizationally sophisticated and to possess strong leadership

In fact in a small number of Jamaican communities gangs have been found to be highly

organized with individual gang leaders being referred to as dons and community leaders The

gang leader in such a community is often found to have substantial control over members and

residents as these communities often turn to the don rather than the police for justice The

don will hold court and punish those who commit crime Punishment can include beatings and

torture as well as execution (Morgensen 2004) Although this level of organizational structure

and sophistication is found only in a small number of Jamaican communities generally the

gangs in Jamaica are believed to have some organizational capacity or at least enough to

reduce violence in communities

Our findings however indicated that prior to 2005 and the death of Neil Wright perhaps only

Jungle 12 could approximate that capacity to discipline members and enforce a truce After the

gangrsquos fragmentation in 2005 Jungle 12 lost much of its organizational capability and

enforcement of the truce was therefore difficult The truce negotiators sought to address the

enforcement issue by proposing a peace council that would involve all parties The proposal

was approved by all key stakeholders still some gang leaders demanded cash payments as a

68

condition for attending council meetings Peace was consistently seen by then as a bargaining

tool rather than as an honest attempt to establish and maintain peace In the end members of

only two gangs were attending the meetings21 and the council soon dissolved

In an effort to replicate the council function UWI sponsored one of the most respected

negotiators a community activist to become a one-person monitoring and intervention

specialist or a ldquoviolence interrupterrdquo His job was to ensure that truce violations did not lead to

a return of the gang wars mdash and there were many violations of the truce For example there

were instances of gang members crossing boundaries and entering the turf of another gang

armed although not initiating conflict behavior that was interpreted by the opposing gangs as

preparation for the next round of ldquowarrdquo or as laying a foundation for a surprise attack that

would exploit the truce for this purpose In the absence of the council these matters were

reported to the violence interrupter who tried to resolve the problems in consultation with the

various gang leaders Often the gang leaders were unresponsive or incapable and therefore the

threatening practices and violence continued Ultimately there were no rules or bodies or

persons who could regulate the violence and there were never any reference points for

compliance The formal truce agreement was an attempt to negotiate and impose such rules

via a collective pressure that would include third parties but it was unsuccessful in doing so

The potential for re-engineering norms related to conflict thus was not realized

Another explanation for the failure of the gang truce might be that it was more a vehicle for

rhetoric rather than for reality The gang leaders insisted that they would sign the truce

agreement only if it were ratified in public with the presence of the media (Jackson 2008 Levy

2009) The leaders might have viewed the process in and of itself as a means of increasing their

reputation and influence within the community and in policymaking circles (and to reduce

mutual distrust) In signing the truce gang leaders publicly pledged to reduce their involvement

in violence thereby calming local residentsrsquo fears They also made public efforts to increase

resources for their communities perhaps in an attempt to portray themselves as ldquoprovidersrdquo to

the community In fact the truce did provide gang leaders with an opportunity to be seen in

21 Interestingly the Jungle 12 factions did not attend any of the peace council meetings

69

public collaborating with important community stakeholders The imagery of the public signing

was of the government (via the PMI) and others approaching the gang to ask them to use their

means of informal social control in the community to reduce violence mdash to accomplish

something that the government could not do on its own As a consequence the process may

have been perceived by gang leaders as a victory because it enhanced the gangsrsquo reputation

with both the government and community

Alternatively from the start the gangs might not have been fully invested in the gang truce

One of the major criticisms of the Greater August Town gang truce was that gangs were not

required to give up their firearms although some believed that this was an unrealistic request

their demand and the demand of many that all guns be turned in immediately was

quite unrealistic given the decades of ingrained gun culture and the continued inability

of the security forces to guarantee protection for any corner against armed rivals It was

obvious to most observers that that kind of situation could not be ended overnight and

that this was a reasonable first step in the process (Levy 2009 63)

The gangs feared that if they were to disarm themselves they would be vulnerable to other

gangs and unable to protect themselves a concern that appears not to have been addressed by

mediators Indeed at times some elements within the community felt somewhat dependent on

the gangs to maintain security If the gangs would have been disarmed and there were no

near-term alternative prospects for any form of social control both the gang and the

community might have faced additional violence as has been observed in the past In the end

the gang truce only called for a reduction in gang violence and did not provide any solutions to

address the larger problems between the gangs nor did it provide the gangs with any tangible

benefits for abiding by the truce

70

Case Study C Gang Trucemdash The Honduran Experience

Introduction

Violence in Honduras is at epidemic levels increasing almost 44 percent over the past five

years In 2012 there were 7172 homicides in Honduras or about 86 homicides per 100000

population (Instituto Universitario de Democracia Paz y Seguridad 2013) making it the most

violent nation in the world (United Nations 2013) Likewise Hondurasrsquo second largest city San

Pedro Sula has the highest municipal level homicide rate in the world with 1290 homicides

(Instituto Universitario de Democracia Paz y Seguridad 2013) or about 174 per 100000

population (United Nations 2013) In comparison the average homicide rate across the globe is

about 62 per 100000 and the average homicide rate in Central America is about 27 per

100000 (United Nations 2013)22

Much of the discussion about the causes of Hondurasrsquo high homicide rate has focused on its

relationship with international drug trafficking routes gangs and conflict between crime

groups and the government and government instability Estimates of gang involvement very

widely but some have suggested that there are between 12000 (Seelke 2012) to 36000

(Ratcliffe et al 2014) gang memberrsquos in Honduras who typically belong to one of two gangs

MS-13 and 18th Street These gangs are said to be less organized than their counter parts in El

Salvador but are said to be just as involved in extortion and intimidation and perhaps more

involved in drug trafficking because of their stronger linkages with Mexican drug cartels

(Wilkinson 2013)

22 An unusual characteristic of the homicide problem in Honduras is the age of victims Typically in the Western Hemisphere homicide victims are aged 15 to 29 In Honduras however those 30 to 44 have the highest rate of violent victimization For example 1 out of 280 males 30 to 44 years old are the victim of homicide compared to 1 out of 360 males 15 to 29 years old (United Nations 2013) These findings by themselves are suggestive of a chronic gang problem (Spergel 1995) Honduras also stands out in the Western Hemisphere in the proportion of its homicides that involve a firearm In 2012 about 84 percent of the homicides involved a firearm (11) The proportion of homicides that involve a firearm appears to be increasing as well In 2008 79 of homicides involved a firearm compared to 81 in 2009 83 in 2010 (United Nations 2013)

71

Over the past decade the nation has responded with ldquoMano Durardquo (ie iron fist or heavy hand)

The new legislation provided the police with more authority to stop search and detain gang

members The new legislation also permitted the courts to sentence gang members to prison

for 12 years for simply being a member of a gang and allowed the courts to sentence

individuals to even longer prison terms for gang related incidents Concomitantly the military

joined the effort to fight gangs by patrolling neighborhoods along side the police While the

public and media strongly supported the shift in national policy toward Mano Dura much of

the evidence suggests that these legislative and policy changes were not effective as the

number of homicides continued to escalate Some suggest that its lack of success was because

gang members who were arrested were released due to of lack of evidence or those who went

to prison if they were not a gang member before entering prison joined a gang Other critics

point out that the heavy handed approach by the government led to loss in the rule of law as

vigilantes engaged in extra-judicial violence against gang members (Seelke 2012) Still others

said that the legislation and policies never really had a chance of working because of the

general lack of effectiveness of the police and courts and the wide spread corruption

throughout the criminal justice system (Zilberg 2011)

As a consequence of the above policymakers and citizens voiced optimism about the possibility

of a truce between gangs after initial results in El Salvador suggested the strategy might be

effective (Villiers-Negroponte 2013) Honduran church leaders and the Organization of

American States (OAS) began to develop a strategy to implement a similar type of truce in

Honduras and the President offered his personal support in their efforts (Arce 2013) In this

case study we examine the processes that lead to the Honduras gang truce and the nationwide

impact of the truce on homicides In the below section we discuss the major stakeholders who

participated in the truce processes leading up to the truce and the establishment of the truce

Key stakeholders

The primary facilitator for the truce process in Honduras was Archbishop Roacutemulo Emiliani who

had earlier served as the Assistant Bishop of the Dioceses of San Pedro Sula and who received

72

support from the Catholic Church to pursue the truce (Bosworth 2013) Prior to the

negotiations he was well known for his work which attempted to establish peace between the

gangs and his advocacy for prison reform and social reintegration programs for gang members

(The Daily Herald 2013) From the onset Monsignor Emiliani proceeded cautiously to ensure

reasonable expectations among the public and policymakers He maintained publically that ldquohe

didnacutet want to be a salesman of false promises about what was going to occur in the future the

things that they do are unpredictable but we expect to have a declaration of reconciliation

principles with societyrdquo (El Mundo 2013) Additionally he wanted to set reasonable

expectations because he knew that it would be a ldquoslow painful and draining processrdquo (El Nuevo

Siglo 2013) and that ldquoWhat is coming is difficult It is not easy It is complicatedrdquo (Castillo

2013)

As in El Salvador the Organization of American States (OAS) played a major role in facilitating

the peace process alongside Monsignor Emiliani Adam Blackwell served as the Secretary of

Multidimensional Security for the OAS and represented Canada on the Honduras Security

Reform Commission (Willcocks 2014) His participation in the mediation process was requested

by Honduran gang members who were in prison They requested that the OAS help broker a

peace agreement with the Honduran government and to help identify resources that would

assist gang members to obtain legitimate jobs (Associated Press 2013) The OAS together with

the Catholic Church served as a ldquobridgerdquo between the executive branch of the government and

the two gangs Additionally two of the mediators (ie Salvadoran Army officer and Police

Chaplain Monsigor Colindres and former Salvadorian congressman Mijango) who helped broker

the truce in El Salvador provided additional support to Monsignor Emiliani and Secretary

Blackwell They traveled to Honduras to present their experiences with the gang truce in El

Salvador and to convey that a gang truce is a promising and legitimate strategy for addressing

gang violence (Associated Press 2013)

Gang leaders of the two primary gangs in Honduras (MS 13 and 18th Street) also participated

extensively in the negotiation process It was stated that they had become weary of the violent

conflict and understood that a truce would be beneficial to the Honduran people (Servellon

73

2013) From the beginning however a number of the critics of the truce argued that Honduran

gangs did not have the capacity to control street level violence They characterized the

Honduran gangs as having less organizational leadership (Bossworth 2013) less control over

turf (LatinNews Daily Report 2013) and being more organizationally fractured (Dudley 2013)

than MS 13 in El Salvador

At the time that discussions about the possibility of a gang truce began President Porfirio Lobo

Sosa was publically supportive of the Catholic Church and OAS negotiating with the gangs

Media reports quoted the President saying I am ldquoprepared to do what ever is necessaryrdquo to

support the mediators (Phillips 2013) ldquoWe have to look for anything thatrsquos an alternative to

violencehellipOn the part of the government we are open to any process that can lower violencerdquo

(Associated Press 2013) and that he had given ldquohis blessing to Emillanirsquos efforts to broker peace

between the gangshelliprdquo (The Daily Herald 2013) However in November 2013 after a general

election the new president Juan Orlando Hernaacutendez through his recently appointed Vice

Minister of Security declared that the government would no longer support the truce process

with the gangs (El Heraldo 2014) Since then the Government of Honduras has not mentioned

the peace process that was initiated in May 2013

Truce making process

It is important to note that prior to the announcement of the gang truce a number of key

stakeholders were somewhat skeptical about its possibility On the one hand some suggested

that a gang truce had been attempted in the past with no success For example one

stakeholder commented to an international media outlet that ldquoEveryone here agrees itrsquos a

positive step forward but people are cautiously optimistic because in 2005 these two gangs

had another peace treaty with each other Now that treaty was very tentative it only lasted

less than two monthsrdquo (Al Jazeera 28 May 2013) On the other hand as noted above other

stakeholders believed that the local gangs did not have enough organizational leadership to

change the behavior of gang members to reduce violence (Bosworth 2013) They argued that

even if gang leaders wanted a gang truce there was no way of enforcing it on the streets

74

Several months prior to the announcement of the gang truce Carlos Mojica Lechuga an 18th

Street Salvadorian gang leader publically stated that representatives of MS13 and 18th Street

in Honduras spoke with several gang truce key stakeholders in El Salvador for the purpose of

replicating the truce in Honduras Tellingly in reflection of the visit Mojica noted that the

advantage of a gang truce is that it formally recognizes Honduran MS13 and 18th Street leaders

as important political persons within the nation He also noted that Honduran gang leaders

have historically been treated poorly and that a gang truce holds the potential for

demonstrating the political power of each of the Honduran gangs (Villiers Negroponte 2013)

The negotiators used different language to describe the early days of the truce Specifically

they mentioned that there had been a consultative process with the gang leaderships so they

were in a process like lighting23 Prior to the truce media sources mentioned that the gang

leaders were offering to stop violence and to not recruit more youth into the gangs24 The gang

leaders also spoke about the ldquopersecutionrdquo they and their family members had suffered during

the previous years highlighting that they had been prohibited social opportunities offered to

others in society (Arce 2013)

Leading up to the negotiations the leadership of both gangs expressed their interest in three

goals 1) lowering violence and crime 2) reconciliation with God society and the government

and 3) helping to improve the social conditions of their communities Although there was not

written documentation on the exact agreement between the parties one MS-13 leader

affirmed that the pact would include all violence (El Comercio 2013) However when talking

about sensitive topics such as extortions which is one of the main sources of income for the

gangs the gang leader said that ldquowould be taken up at a later daterdquo (ABC Internacional 2013)

Leaders of 18th street made similar general statements about ending violence but they were

more specific about their demands One of the 18th Street leaders stated that ldquowhat we want is

23 El Universal 2013 - httpwwweluniversalcominternacional130531obispo-hondureno-descarta-tregua-entre-las-pandillas 24 (Garcia 2013 -httpwwwlaprensahncspmediapoolsitesLaPrensaHondurasSanPedroSulastorycspcid=338546ampsid=276ampfid=98-)

75

to have a dialog with any commission appointed by President Porfirio Lobo and we are sure

that the situation in Honduras will begin to changerdquo (El Nuevo Siglo 2013)

Establishing the gang truce

On May 28th 2013 with public declarations from leaders of both gangs the gang truce was

announced From the beginning of the process the role of the government in the truce was

unclear (La Prensa 2013) Likewise there was little discussion about the exact nature of the

agreement the terms in which gang members would abide and any benefits that would be

made available to those who participated in the truce For example as one stakeholder

indicated the government never decisively considered viable proposals to give the members of

the gangs any opportunities It is important to note that none of the parties signed a formal

commitment and neither MS-13 nor 18th Street signed any type of ceasefire agreement The

gang leaders were in separate locations and were never in direct contact with each other during

the announcement That is both gangs seemingly agreed to the gang truce without ever

talking to each other The ldquopeace processrdquo was publicly announced on May 28 2013 through

ldquojoint but separaterdquo declarations made by the leaders from both gangs imprisoned in the San

Pedro Sula prison (National Penitentiary SPS) (The Daily Herald 2013) The national and

international media widely covered the declarations

The MS-13 leaders said they would not commit any more homicides or any other types of

crimes They ensured that this was an ldquoimmediaterdquo order and would be effective throughout

the country They emphasized ldquoall of the boys know what they have to do starting todayrdquo

(Pachico 2013) The leadership of 18th Street declared that they would stop violence and other

criminal activities but also indicated that the government would have to ldquolisten to themrdquo

Little research has examined whether the gang truce in Honduras ever impacted violence in the

nation Instead anecdotes have been used to portray its effectiveness One facilitator for

example indicated ldquoin Honduras the dialog with the gangs has been positive however the sad

thing in Honduras is that the two main gangs have not accepted a truce between them they

just haven`t accepted it as yetrdquo (La Prensa 2013) despite the fact that ndash in their own words ndash

76

ldquothey do want to hold a dialog with society with the government and with the policerdquo

Conversely gang leaders declared that the truce had been effective For example a leader of

18th Street noted that ldquohellipIt has already done its part telling members in the areas the gang

controls to stop the violence and crimehellip[estimating] crime had already dropped 80 percent in

those areasrdquo (Associated Press June 17 2013) Similarly a member of MS13 estimated that

violence in MS13 controlled areas declined by 45 percent (Associated Press June 17 2013) As

a symbolic gesture of the impact of the truce MS13 leaders also noted that as a gesture of good

will they made and delivered 60 beds for a nursery home in San Pedro Sula (Associated Press

June 17 2013)

Methods

For the present case study we used a pre-test post-test longitudinal quasi experimental design

Data from the 2001 Honduran Census was obtained from the National Institute of Statistics

(INE) These data provided municipal level measures of number of residents population density

per kilometer percent of population who moved in from another municipality ethnicity

percent urban number of residents immigrated to the United States percent female headed

households percent unemployed age composition income percent of households rented and

education level In addition population projections for the years 2005 2010 and 2014 were

also obtained from the National Institute of Statistics (INE) The population levels for the total

population as well as the percentage of residents in a municipality that rural and percentage of

residents who are female were linearly interpolated for the intervening years Examination of

the observed levels indicated that growth was linear overall and so we feel confident that our

linear interpolations are good approximations Second we used municipal level homicide data

by month and year for the period May 2012 through July 2014 These data were provided by

the Honduran National Police through the Honduran US Embassy Both datasets were

merged for the present analysis

Measures

77

The dependent variable for the Honduran case study is the monthly homicide rate which was

calculated by dividing the number of homicides in each municipality by its population and

multiplying this figure by 100000 We examined change by comparing the homicide data 13

months prior to the gang truce to the homicide data 14 months following the truce in each of

the nationrsquos 298 municipalities As presented in Exhibit 21 there were a total of 7910 homicides

over the study period with each municipality averaging 183 homicides (sd=837) A trend

analysis showing the monthly number of homicides on a national level prior to and following

the gang truce is presented in Exhibit 22 It shows that nationally the homicide rate gradually

declined over the study period

Exhibit 21 Summary Statistics

Pre-Truce (n=3809) Post-Truce (n=4101) Total (n=7910)

Mean SD Mean SD Mean SD

Homicide 200 900 166 773 183 837

Population 2874492 8674659 2939557 8870267 2908226 8776124

Homicide rate 521 953 413 814 465 885

Density 9926 10878 9928 10878 9927 10877

Born in the same municipality 8211 1286 8211 1286 8211 1286

Other population group 8545 2390 8545 239 8545 2390

Percent rural 7926 2732 7899 2762 7912 2748

Living in another country 331 650 331 65 331 650

Socioeconomic status 005 097 -002 102 001 100

Exhibit 22 Homicide rate in Honduras by month

78

We used census data to control for several community-level structural factors Principal

components analysis was used to reduce some of the data into a summary measure Exhibit 23

presents the results of the component loadings One component was extracted that we labeled

socio-economic status (SES) This component exhibited high loadings for percent female

headed household percent unemployment and percent completing primary education

Population was used as our exposure variable and population density (per 1 km) residential

stability percentage of the population that is rural percentage of the population that is

indigenous (verify) and the number of residents immigrated to the United States served as our

control variables Population density and outmigration were logged to address skewness in

these two measures

Exhibit 23 Factor Analysis for Socioeconomic Status

Eigenvalue 236093

Variables Factor Loadings

Female-Headed Households 07977

Unemployed 09269

Primary Education 09303

Findings

The results of our t-test analysis are shown in Exhibit 24 It shows that there was a significant

decline in the homicide rate at the municipal level before and after the gang truce The

homicide rate prior to the truce was 697 per 100000 population and following the truce the

homicide rate was 566 homicides per 100000

79

Exhibit 24 T-test on National Data by Month (n= 27 Months)

Homicide Rate Pre Truce Post Truce Change

Mean 6972346 5663592 -1308754

SE 01677361 01151668 02009224

plt 001

Next we examined the effect of the truce through ARIMA regressions The first model in Exhibit

25 does not include an effect of the temporal trend In this model we again find that homicides

declined significantly in the period following the truce We then employed the ARIMA model

again but this time included a variable (month) to control for the temporal trends in the data

Model 2 in Exhibit 25 shows that when we controlled for temporal trends the impact of the

truce we observed was no longer significant Our findings suggest that homicide did not decline

as a consequence of the gang truce but instead the decline in homicides was part of a longer

term historical trend that was independent of the truce

Exhibit 25 Results from the ARIMA Models (n= 27 Months)

Homicide Rate Model 1 Model 2

Effect SE Sig Effect SE Sig

Truce -1309 0350 -0906 0595

m2 -0623 0481

m3 -0153 1447

m4 -0074 0607

m5 -0104 0450

m6 -0050 0466

m7 -0002 0516

m8 0003 0404

m9 0024 1633

m10 -0396 0522

m11 -0754 2424

m12 0837 0366

Time -0030 0047

Intercept 6973 0254 7258 0472

Autocorrelation Coefficient 0010 0408 0111 0249

plt 001

80

Last we used a fixed effects multi-level negative binomial regression to predict the homicide

rate with population as an exposure variable and controlled for the socio-demographic

characteristics of each community The results are presented in Exhibit 26 The only significant

variable in the analysis was the impact of municipal level population stability Specifically we

found that homicide rates increased in areas where residential mobility was high Once again

our analysis showed that time had a negative effect indicating that homicides were decreasing

in all areas over the study period The gang truce itself was unrelated to the decline in

homicides

Exhibit 26 Results from the Multilevel Negative Binomial Models (n= 27 Months)

Homicide Rate Model 1 Model 2

Effect (SE) Sig Effect (SE) Sig

Fixed Effects

Homicide Rate 0004 0004

(0002)

Truce -0119 -0129

(0073) (0072)

Time -0008 -0007

(0005) (0005)

Month 1 0024 0024

(0059) (0059)

Month 2 -0107 -0107

(0062) (0062)

Month 3 -0041 -0041

(0063) (0063)

Month 4 -0033 -0033

(0065) (0065)

Month 5 -0002 -0002

(0067) (0067)

Month 6 (reference)

Month 7 0027 0027

(0051) (0051)

Month 8 0023 0023

(0057) (0057)

Month 9 0036 0036

81

(0056) (0056)

Month 10 -0051 -0051

(0057) (0057)

Month 11 -0031 -0031

(0058) (0058)

Month 12 0187 0187

(0056) (0056)

Born in the same municipality -0013

(0003)

Pecent rural -0003

(0002)

Socioeconomic status 0029

(0044)

Percent dominant population 0001

(0002)

Density (ln) -0010

(0050)

Living in another country (ln) 0001

(0061)

Intercept -10042 -10051

(0061) (0058)

ln(Alpha) -2368 -2367

(0148) (0148)

Random Effects

Var(Truce Effect) 0066 0066

(0024) (0025)

Var(Intercept) 0388 0309

(0049) (0042)

Cov(truce effect intercept) -0024 -0006

(0027) (0025)

plt 001

82

Conclusions

Violent crime in Honduras is widespread With a homicide rate of about 86 per 100000

population Honduras is one of the most violent nations in the world (United Nations 2013)

This compared to an average homicide rate around the world of 62 per 100000 people and

about 27 per 100000 in Central America (United Nations 2013) Given the high rate of violence

in Honduras it is clear that new and innovative ways of reducing violence should be considered

This case study gave an overview of the implementation and impact of the gang truce

experience in Honduras in 2013

The goal of the truce was to significantly reduce the number of homicides In order to assess

the impact of the gang truce in Honduras a pre-test post-test longitudinal quasi-experimental

design was used Census and homicide data were merged at the municipal level to asses

whether the truce had an impact on homicides controlling for population characteristics and

the natural trend in violent crime

Overall the findings suggest that while the homicide rate in Honduras was on a slight

downward trend the gang truce itself was unrelated to any homicide reductions That is the

gang truce had no measurable impact on homicides in Honduras Given the drastic reductions

achieved in El Salvador and the fact that Hondurasrsquo gang truce was a replication of El

Salvadorrsquos the following discussion will examine two important differences between the two

countries gang truces First the implementation of the truce in Honduras did not appear to

obtain trust between all parties involved and did not achieve any notable short-term

deliverables That is the implementation was not robust Second some suggest that the gangs

in Honduras do not have the organizational sophistication to be able to control their members

on the streets rendering them incapable of carrying out any truce agreements

First the implementation of the truce in Honduras did not appear robust The communication

between the parties was weak and none of the parties completed any significant actions as part

of the truce The Catholic Church and OAS served as the bridge between the two main gangs in

Honduras MS13 and 18th Street and the executive branch of the government The ldquopeace

83

processrdquo as it was called in Honduras was initiated in May 2013 After the general election in

November 2013 the newly appointed administration declared that it would no longer support

the truce process with the gangs As a result the truce was short-lived and the governmental

support for the effort shifted with the change in leadership

Though the negotiations were largely based on the good will of a well-respected negotiator

few tangible incentives were offered during the process The negotiations from the beginning

involved discussions about large scale social programming Given the short time frame these

goals in hindsight were unrealistic Naturally there was little trust between the parties at the

beginning of the negotiations and without any quick tangible deliverables from either side the

truce never really materialized

The second challenge to implementing a successful gang truce in Honduras might have been

related to the nature of the gangs themselves It was unclear whether the gangs possessed the

level of cohesion and hierarchical leadership required to make some measures feasible If the

gangs do not have the organizational capacity to control their members on the streets any

agreements that come from the negotiations would be difficult to implement In the earlier

case study of the Salvadorian experience it was suggested that the successes in that country

were related to one of the gangrsquos organizational capacity to impose control of its members The

gangs in Honduras might be different There is at least some evidence that MS 13 in Honduras

might have less organizational leadership (Bossworth 2013) less control over turf (Latin News

Daily Report 2013) and have been more organizationally fractured (Dudley 2013) than their

counterparts in El Salvador The gangrsquos ability to operate as an efficient organization can greatly

impact the outcome of the truce process In general we know that gangs do not have very high

levels of organizational sophistication (Decker Katz and Webb 2008 Decker Bynum Weisel

1998) It might be that gangs in Honduras are more the norm in terms of organizational

capacity

In summary the 2012 truce negotiations in Honduras did not produce any measurable

reductions in the homicide rate The gangs wanted to speak to Honduran society and they even

84

preferred to speak with the governmental authorities but they never interacted with either

The negotiation process seemed to end as quickly as it started The gangs did not deliver with

lower rates of violence and the government did not provide social programs It would be safe to

say that a robust gang truce did not materialize in Honduras The implementation of the truce

seemed to struggle for two primary reasons First the mediators were not able to accomplish

any quick wins to build trust between the parties involved The commitment level on all sides

was not clear throughout the process Second it is not clear whether the gangs in Honduras

have the organizational capacity to control members on the street as would be required to

carry out an effective gang truce In short the 2013 gang truce in Honduras was unsuccessful

85

Conclusions Policy Implications and Recommendations

The purpose of this report was to systematically examine the effectiveness of gang truces Gang

truces have been widely implemented but rarely evaluated Of those gang truces that have

been evaluated little attention has been given to why and how they came into existence In

this report we reviewed prior research on gang truces and presented case studies of gang

truces implemented in El Salvador Jamaica and Honduras for the purpose of understanding

the negotiation processes undertaken with and between gangs and other stakeholders We

were interested in identifying the actors involved in the negotiations the goal(s) of the

negotiations and the strategies employed to carry them out Most importantly however we

wanted to determine whether the gang truce resulted in a reduction in the number of

homicides Each case study offers lessons learned that are unique to their particular

circumstances and when considered together provide direction to policymakers on the

benefits and risks of implementing gang truces

The case studies presented here constitute the most comprehensive evaluations of gang truces

to date Existing documents were used to collect information about the processes associated

with each gang truce Many of these documents included such items as peer reviewed articles

books and reports The majority of these documents were collected over the Internet

requesting documents from those close to the truce and searching library databases Related

the case studies made use of articles obtained from local newspapers The newspaper articles

were not only intended to provide a historical record of the development of each gang truce

but also to provide additional insight into the various external forces that might have impacted

the gang negotiations Because the newspaper serves as a forum for the community to speak

about its concerns newspaper articles also provided a rich source of data on how those in the

community felt about the gang truce Accordingly the newspaper articles offered a different

view of the problem and offered different opinions as to how a gang truce should or should not

be implemented We also conducted a small number of in-depth qualitative interviews with

key informants These data were collected to supplement existing documents and to clarify

issues associated with the negotiation processes This included but was not limited to

86

questions pertaining to identifying the actors involved in the negotiations the goal(s) of the

negotiations and strategies employed to carry out negotiations The interviews were intended

to obtain information from those who possessed first hand knowledge about the gang truce in

each nation

We examined the impact of each truce using official data We first performed a simple t-test

comparing the homicide rates before and after the truce However as discussed above this

technique has limitations The most severe of which is that even if the test was significant it

would be difficult to determine whether the difference was due to the gang truce or a natural

change over time in the outcome We addressed this limitation by using time series models

whereby the homicide rate for the community was modeled as a function of time with truce

period indicators included to measure the effect of the truce net of the temporal trends These

models were estimated with ARIMA techniques Supplemental models were also employed to

examine and control for factors other than the truce that might have impacted homicide over

the study period

Summary of Findings Related to the Implementation of a Gang Truce in the Three Sites

We found that the implementation of gang truces have a number of common characteristics

The first is that in each case a community was experiencing an uncharacteristically high number

of gang related homicides over a fairly lengthy period of time The continued high level of

violence in each case resulted in the community placing strong pressure on the government in

general and the justice system in particular to respond to the problem quickly and effectively

In each case they had first attempted to control gang violence through suppression oriented

strategies and these strategies were found to be ineffective over the intermediate and long

term In turn each communityrsquos inability to exercise traditional informal and formal social

control to decrease levels of violence became self evident to the public and government This

resulted in both the state and community to seek (or participate in) an alternative strategy in

which negotiators would formally andor informally work with gang leaders to establish a truce

that would reduce gang homicide

87

Key stakeholders involved in the negotiation and establishment of each gang truce were fairly

similar In each of the cases examined the gang leaders of the largest and most violently

involved gangs were willing to consider participating in negotiations that could lead to a truce

In each of these cases it was clear that the gangs not only sought to collaborate with the

negotiators for the purpose of reducing violence but perhaps more importantly were seeking a

means in which to gain greater more positive recognition in the community and to reap some

form of benefit to themselves their members and possibly their community In each case

while not always formally involved government officials were at a minimum made aware of

negotiations and in some cases solicited the assistance of third partyrsquos to broker an agreement

between stakeholders In each case it was at least implicitly understood that the government

would ldquolistenrdquo to the gang leaderrsquos expectations of the government and what theymdashthe gang

leaders--had to offer in exchange We found that when the government was no longer willing

to ldquolistenrdquo to or collaborate with negotiators the truce processes ended abruptly Negotiators

were typically comprised of a very small group (ie 2-3) of individuals who were perceived to

be ldquohonest brokersrdquo In El Salvador and Honduras this included a high ranking Catholic Church

official a leader from an international diplomatic organization (ie OAS) and other neutral

parties In Jamaica this included a quasi-governmental organization that had been established

for the purpose of brokering negotiations between gangs to reduce violence and the local

university which had access to staff who were perceived to be neutral but had an interest in

reducing violence due to its proximity to the university

The strategies used to execute each gang truce were similar but yet importantly different They

were similar in that each involved a team of negotiators working to identify common goals to

be achieved and identifying tangibles that could be delivered to the gang leaders gang

members and their community in exchange for the gang achieving their stated goals They

were different however in terms of the structure of the delivery of each parties promise to the

other In Honduras and Jamaica it appears that gang leaders committed to reducing gang

violence in exchange for general promises made by the negotiators for example that

substantial public works programs would be implemented for the goal of reducing

unemployment among gang members and the community In both of these cases it required

88

the government to develop and deploy large scale social programming in a very quick period of

timemdashsomething that neither government had a strong record of demonstrating In El Salvador

negotiators employed a strategy of the gang leaders promising to deliver immediate changes in

gang member behavior for immediate administratively natured changes by the government

For example in exchange for a reduction in gang violence the government agreed to relocate

imprisoned gang leaders to less restrictive prisons and provide them some privileges Following

the successful execution of the first part of the truce which resulted in near term success for

both parties they began to negotiate issues that would take a longer period of time for the

gangs and government to deliver Our findings suggested that some promised deliverables need

to be easily and quickly delivered early in the process so that trust increases between both

parties Stakeholders only have a brief period of time to provide promised benefits before trust

is lost and that tangible benefits need to be delivered in weeks or months not years

Summary of Findings Related to the Impact of Gang Truces in the Three Sites

Our findings suggest that El Salvadorrsquos gang truce had a significant and dramatic impact on the

homicide rate The mean number of monthly homicides declined from about 354 prior to the

truce to about 218 following the truce for a net decrease of about 136 homicides per month

Our forecast showed that between March 2012 and June 2014 the truce had saved about 5501

lives However there was no significant change between the pre-truce and post-truce periods

in the number of thefts extortions robberies rapes and auto theftsrobberies We also found

that the gang truce did not result in a homogenous decline in violence across municipalities

About 61 percent of municipalities experienced a decline in homicides but the decline in

violence varied substantially between municipalities We examined this issue further by parsing

out the relative influence of the number of MS13 and 18th Street gang members on the street

and in prison from each municipality Our analyses indicated that following the truce the

number of MS13 and 18th Street gang members on the street in a municipality was not

significantly related to a decline in homicide but the number of imprisoned MS13 and 18th

Street gang members was associated with a significant change in homicides following the gang

truce In particular the number of imprisoned MS13 gang members was associated with a

89

significant decline in homicides following the gang truce and the number of imprisoned 18th

Street members was associated with a significant increase in homicides following the truce

In Jamaica our initial findings showed that the gang truce might be an effective mechanism for

reducing violence Bivariate analyses showed a significant decline in homicides immediately

after the truce was implemented This explains the work previously published by policymakers

researchers and news reporters Upon further examination of the data however comparing

change in the target and comparison areas and accounting for temporal trends we found that

the decline in homicide was part of a larger nationwide decline in violence and that the gang

truce was not responsible for the decline The only significant effect that we uncovered was the

possibility that homicides were displaced outside the target area for a brief period of time but

then returned to normal

Our findings from Honduras told a similar story as Jamaica Initial analysis showed that the

number of homicides on average declined across municipalities following the gang truce

Specifically the mean number of homicides declined by 13 per 100000 population with 687

homicides per 100000 population on average occurring in each municipality prior to the truce

and 566 homicides per 100000 population on average occurring in each municipality after

the truce However after we examined the effect of the truce through the ARIMA model and

included a variable (month) to control for the temporal trends in the data the impact of the

truce we observed in our bivariate analysis was no longer significant Our findings as in

Jamaica suggested that the decline in homicides was not the consequence of the gang truce

but instead the decline in homicides was part of a long term decline in homicides due to

exogenous factors

The Potential Benefits and Consequences of a Gang Truce

Over the last several years there have been a number of naturally occurring experiments

involving gang truces in a variety of nations in various regions of the world Findings from

evaluations of gang truces are mixed As noted above in El Salvador the gang truce could be

characterized as highly effective at least for the two years following the truce It is worth

mentioning that even after the truce breakup homicides rates while above truce levels

90

continue slightly below pre-truce levels In Jamaica and Honduras the gang truce had no impact

on violence In Los Angeles and Trinidad there was evidence that violence decreased for at

least ninety days but then increased substantially beyond those rates observed prior to the

gang truce (see the introduction section of this report for this discussion) As a consequence it

appears that the potential for long term consequences might out weigh the potential for short

term benefits Only one study site(El Salvador) demonstrated a truce having a substantial and

long term impact on violence Others conversely demonstrated the truce had no impact or

increased violence over the long term In fact a number of scholars have noted that gang truces

are likely to result in a boomerang effect with gang violence increasing over the long run

because of enhanced cohesion within the gang (Klein 1995) Maguire (2013) notes that when

government officials negotiate a truce with gangs they might ldquoinadvertently be acknowledging

gangs as legitimate social entitiesrdquo (p 11) This in itself might increase cohesion among gangs

which has been found to be associated with increased levels of criminality (Decker et al 2008

Klein 1971 Maguire 2013)

Further research is needed to examine how gang truces might impact group cohesion and if it

does whether this in turn results in greater violence Gang truces convey the well-intentioned

image that violence has been addressed and policymakers are doing something about the

problem but researchers need to better understand the probability of a gang truce reducing

violence increasing violence or having no impact This will better position policymakers to

understand the relative risks associated with these types of interventions

Our findings also suggest that while gang truces could be an effective intervention in areas

where gangs are highly structured and organized such as El Salvador they could be counter-

productive in areas where gangs are not as structured and organized Because the vast majority

of street gangs are not well organized (Klein 1995 Spergel 1995) the utility of a gang truce in

reducing violence might be limited Our findings coupled with prior research suggest that gang

interventions need to be tailored to the nature of the gang and its members or it risks

increasing gang violence

Final Thoughts

91

Our analysis suggests that gang truces should only be used as a means of last resort and then

only under certain conditions Given the risks associated with a gang truce communities with

high levels or at least modest levels of formal social control should rely on other more

promising gang control strategies Only when the state has limited or greatly reduced capacity

for social control should a truce be considered Concomitantly a gang truce should only be

considered when a community is experiencing a substantial amounts of gang violence

Communities that are experiencing minimal to modest amount of gang violence may have more

to lose from the establishment of a gang truce than they have to gain Additionally our

findings suggest that a gang truce might only be feasible when gangs are sufficiently organized

to the extent that they have the capacity to regulate memberrsquos behavior In other words gang

leaders must have the ability to reduce their memberrsquos involvement in violence for a gang truce

to work Our findings more concisely suggest that gang truces should only be considered when

there are a great number of gang homicides the state has limited capacity to address the

problem and gang leaders have enough informal social control over their members that they

themselves can substantially control the levels of violence in their community by regulating

their memberrsquos behavior

Recommendations

Gang truces are conjunctural strategies States who suffer from gang-related violence must

establish permanent public policies for crime control and prevention A government that

considers implementing a gang truce should be aware that it cannot become the center

strategy of its public policy for citizen safety

Gang truces should only be used as a means of last resort and then only under certain

conditions Stakeholders must determine whether a process of dialog or negotiation with gangs

is legal ethical and feasible

Stakeholders must anticipate demands that are likely to arise and their response options Some

demands may be easily met such as improved prison conditions Others are much more

difficult and amorphous such as community development through more integrated violence

92

prevention programs (such as those implemented by SolucionES in El Salvador) local economic

development programs or economic reinsertion of ex-gang members

Stakeholders should incorporate immediately achievable and demonstrable deliverables Long-

term goals and promises are unlikely to create the trust needed to sustain a gang truce

Stakeholders must first determine the position in which they are negotiating the incentives that

are possible to deliver and the boundaries and limits they face Gangs are mostly likely to trust

representatives from NGOs community-based organizations and members of the faith based

community as brokers because they are considered more reliably neutral advocates for peace

They need to understand the capacity of the government to deliver promises in a timely

manner

Governments have to make a choice about the visibility and transparency of its participation

This decision needs to be made in the context of the national and local laws the publicrsquos

expectations of transparency and patterns of practices of the past

Governments must be strategic in their support for truce initiatives Some donor funded

programs run by the government prohibit gang member participation and if the government

does not receive approval from the donor it may risk the donor withdrawing its sponsorship of

the program

Governments must ensure an inter-institutional coordination for the management of truces to

avoid the responsibility to be of a single government institution It is necessary to generate or

collect reliable and pertinent data that can be used to analyze and assess the process

It is necessary to implement an effective monitoring system of the truce process similar to that

used in the full report as it can help parties understand what is working and who is delivering

on their promises More specifically the monitoring and truce management system should be

able to identify truce violations and be prepared to respond through the use of legal and

effective practices if stakeholders do not comply

93

Finally it is necessary to develop evaluations of gang truces and monitoring programs and

support violence prevention activities local economic development activities and pilot

programs to support the reinsertion of ex-gang members into society Clearly national

governments municipal governments NGOs and community-based organizations need

increased capacity and resources to discourage the growth of gangs among at-risk youth It is

therefore increasingly important to create economic opportunities for gang members willing to

leave the gangs and find other legal employment Developing and sustaining those

opportunities in nations with high incidences of poverty will require significant international

funding

94

References

Anderson Elijah 1999 Code of the Street New York Norton

mdash 1998 ldquoThe Social Ecology of Youth Violencerdquo In Crime and Justice vol 24 edited by M

Tonry 65-104

mdash 1990 Street Wise Chicago University of Chicago Press

Arce A (2013 May 28) Honduras gangs declare truce ask talks with govt Associated Press

The Big Story Retrieved from

httpbigstoryaporgarticlehonduras-gangs-declare-truce-seek-talk-govt

Archibold Randal 2012 ldquoGangsrsquo Truce Buys El Salvador a Tenuous Peacerdquo The New York Times

Originally published online on August 27 2012

Axelrod R (1986) ldquoAn Evolutionary Approach to Normsrdquo American political science review

80(04) 1095-1111

Ayala Edgard 2012 Gangs Back Plan for Violence-Free Districts in El Salvador

httpwwwipsnewsnet201212gangs-back-plan-for-violence-free-districts-in-el-

salvador Found on November 4 2014

Bakrania Shivit 2013 Policy Responses to Criminal Violence in Latin America and the

Caribbean GSDRC Helpdesk Research Report 934 Birmingham UK GSDRC University

of Birmingham

Bargent James 2013 ldquoRise in Disappearances Feeds Doubts Over El Salvador Trucerdquo August 1

2013 Found at httpwwwinsightcrimeorgnews-briefsrise-in-disappearances-in-el-

salvador-feeds-doubts-over-truce on September 1 2014

Battin Sara R Karl G Hill Robert D Abbot Richard F Catalano and J David Hawkins 1998

ldquoThe Contribution of Gang Membership to Delinquency Beyond Delinquent Friendsrdquo

Criminology 36 93-115

Block Richard 2000 ldquoGang Activity and Overall Levels Of Crimerdquo Journal of Quantitative

Criminology 16 (3) 369-83

i

EXECUTIVE SUMMARY

Background

While there is much literature describing the assumptions issues and effectiveness of crime

suppression (eg Decker 2003 Decker and Reed 2002 Katz and Webb 2006 McCorkle and

Miethe 2002) and prevention strategies (Esbensen and Osgood 1997) much less attention has

been paid to gang intervention programming particularly gang truces Little is known about

how often gang truces occur what conditions give rise to them the role of third parties in

brokering them their transformative effects and their effectiveness In this policy brief

sponsored by SolucionES1 and conducted by FUNDE a member of the SolucionES Alliance with

Arizona State University we systematically evaluate gang truces including reviewing prior

research and presenting evidence on the effectiveness of gang truces that have been

implemented in El Salvador Honduras and Jamaica for the purpose of identifying lessons

learned should other governments or donors wish to support gang truces in these or other

countries

Summary of Findings

We found that the truce in El Salvador resulted in a reduction in homicides that was not the

product of other trends or temporal factors Thus the truce in this country had a short-term

effect in reducing the lethality of violent crime though not necessarily in other dimensions of

insecurity By contrast the truces in Jamaica and Honduras resulted in no impact on violence

The Jamaican and Honduran experiences therefore mirror the results of prior gang truces that

1 SolucionES is multifaceted violence prevention program being implemented by an Alliance of five leading Salvadoran non-profit organizations who have come together to prevent crime and violence The Alliance members are Fundacioacuten Nacional para el Desarrollo (FUNDE) Fundacioacuten Salvadorentildea para la Salud y el Desarrollo Humano (FUSAL) Fundacioacuten Crisaacutelida (known locally as Glasswing) Fundacioacuten Salvadorentildea para el Desarrollo Econoacutemico y Social (FUSADES) and Fundacioacuten Empresarial para el Desarrollo Educativo (FEPADE) with partial funding from USAIDEl Salvador Together these five organizations have widely-recognized expertise in education health community development economic development research and youth leadership they are bringing their combined synergy and strengths to prevent crime and violence in El Salvador SolucionES is using a three-pronged strategy to prevent crime and violence in El Salvador 1) Strengthen municipal crime and violence prevention capabilities and actions 2) Increase social investment by the private sector to prevent crime and violence and 3) Research publish and disseminate findings to inform decision-making on crime and violence prevention This is one of 10 planned policy-oriented studies The full study is available on request

ii

have been studied to various extents including those in Los Angeles and Trinidad and Tobago

Important differences in how the various truces were negotiated may explain the different

results and one important difference ndash the ability of government and non-gang community

stakeholders to promise and immediately produce measurable deliverables ndash appear to be

especially significant

Prior Evaluations of Gang Truces

Little research to date has examined the effectiveness of gang truces and much of the research

that does exist has been restricted to the field of public health Studies of a gang truce in South

Central Los Angeles between the Crips and Bloods by Cotton (1992) and Ordog et al (1993

1995) found temporary reductions in the number of homicides and gunshot wounds during the

truce but these studies failed to point out that despite about a 35 percent decrease in

homicides for the first three months of the truce

homicides then doubled in months four through

eleven compared with the pre-truce period

Similar findings were reported in Trinidad and

Tobago (Maguire Katz and Wilson 2013) where

violence declined for a brief period of time

(again for about three months) but then

increased substantially over the long term (12

months) These results suggest that gang truces

may produce short-term benefits yet result in

long-term adverse consequences

Additionally these studies have suffered from a

variety of flaws including 1) poor theoretical

assumptions about gangs including often

ignoring that they might have the organizational

structure and internal discipline to enforce their

truces 2) relatively weak evaluation designs that fail to account for other potential causes for

What is a Gang Truce

A gang truce is a nonviolent resolution to a larger conflict between groups that has an impact on general levels of violence and other forms of criminality within a community (Ordog et al 1993 1995 Whitehill et al 2012) It differs from conflict interruption resolution or mediation efforts which seek to rapidly intervene typically through outreach workers or violence interruption specialists in episodic violent events between groups in a community Gang truces often involve dialogue and negotiations between multiple parties (eg gangs government NGOrsquos religious organizations) that seek to recalibrate the norms of conflict within and between groups for the purpose of reducing or eliminating violence and other crime

iii

or broader trends related to violence reductions 3) a failure to examine the processes involved

in creating gang truces 4) a failure to examine unintended consequences of the truces beyond

their impact on gun violence and 5) a lack of sophisticated statistical analysis capable of

discerning overall trends cyclic patterns outliers and turning points

Overview of Study Design

This policy brief examines gang truces implemented in El Salvador Jamaica and Honduras

including the processes undertaken with and between gangs and other stakeholders We

collected information about the processes associated with each gang truce from a variety of

sources including peer reviewed articles books reports local newspaper articles and

interviews of persons with first-hand knowledge about the gang truce in each nation We

examined the impact of each truce using official data We first performed a simple t-test

comparing the homicide rates before and after the truce We then employed time series

models not used in other studies to evaluate homicide rates as a function of time with truce

period indicators included to measure the effect of the truce net of the temporal trends We

also employed supplemental models to examine and control for factors other than the truce

that might have affected homicides over the study period

Research Findings

Implementing a Gang Truce

The gang truces studied have a number of common characteristics First in each case a

community was experiencing an uncharacteristically high number of gang related homicides

over a fairly lengthy period of time which resulted in each community placing strong pressure

on the government in general and the justice system in particular to respond to the problem

quickly and effectively Second each community had first attempted unsuccessfully to control

gang violence through suppression-oriented strategies Third each communityrsquos inability to

exercise traditional informal and formal social control to decrease levels of violence became

self-evident to the public and government This resulted in all involved stakeholders wanting to

seek (or participate in) an alternative strategy in which brokers would formally andor

informally work directly with gang leaders to establish a truce that would reduce homicides

iv

Key stakeholders involved in the negotiation and establishment of each gang truce were also

somewhat similar In each case examined the leaders of the largest and most violently

involved gangs were willing to participate In each of these cases it was clear that the gangs

sought to collaborate with the brokers not only for the purpose of reducing violence but

perhaps more importantly as a means to gain greater more positive recognition in the

community and to reap some form of benefit to themselves their members and perhaps their

community In each case while not always formally involved government officials were at a

minimum made aware of negotiations and in some cases solicited the assistance of third

parties to broker an agreement between stakeholders In each case it was at least implicitly

understood that the government would ldquolistenrdquo to the gang leaderrsquos expectations and offers

We found that when the government was no longer willing to ldquolistenrdquo to or collaborate with

brokers the truce processes ended abruptly Brokers were typically comprised of a very small

group (ie 2-3) of individuals who were perceived to be ldquohonest brokersrdquo In El Salvador and

Honduras this included a high ranking Catholic Church official a leader from an international

diplomatic organization (ie OAS) and other neutral parties In Jamaica this included a quasi-

governmental organization that had been established for the purpose of brokering negotiations

between gangs for the purpose of reducing violence as well as the local university which had

access to staff who were perceived to be neutral but had an interest in reducing violence near

the university

The strategies used to execute each gang truce were generally similar but had important

differences Each involved a team of brokers working to identify common goals to be achieved

and tangibles that could be delivered to the gang in exchange for the gang achieving the stated

goals They differed by how each partyrsquos promise was delivered In Honduras and Jamaica gang

leaders traded violence reduction pledges for long-term government and social changes such

as the development of substantial public works programs to reduce unemployment In

Honduras and Jamaica the government was asked to develop and deploy large scale social

programming in a short time framemdashsomething for which neither country had a strong track

record In El Salvador by contrast brokers secured promises for immediate changes in gang

behavior in exchange for feasible immediate deliverables from the government For example in

v

exchange for a reduction in gang violence the government agreed to immediately relocate

imprisoned gang leaders to less restrictive prisons and provide them some privileges Following

the successful execution of the first part of the Salvadoran truce which resulted in near term

success for both parties they began to negotiate broader issues that would take longer for both

sides to deliver Our findings suggest that some promised deliverables need to be easily and

quickly delivered by both parties early in the process to achieve trust and serve as a first test

of gang leadersrsquo ability to deliver Stakeholders have only a brief period of time to provide

promised benefits before trust is lost meaning that tangible benefits need to be delivered in

weeks or months not years

The Impact of a Gang Truce

El Salvador We found that El

Salvadorrsquos gang truce had a definite

impact on the homicide rate The

mean number of monthly homicides

declined from about 354 prior to the

truce to about 218 following the

truce for a net decrease of about

136 homicides per month Our data

show that between March 2012 and

June 2014 the truce had saved

about 5501 lives (see exhibit 1)

From a hypothetical stand it is

possible to make the assumption that a number of these deaths averted could have been

transformed in disappearances and therefore they were not counted within the official

homicide statistics However the results from the analysis point out that the number of

disappearances was not significantly related with the change in the global behavior of

homicides Additionally over the period of analysis (January 2010 to June 2014) there was no

significant change between the pre-truce and post-truce periods in the number of thefts

extortions robberies rapes and auto theftsrobberies

Exhibit 1 Forecast of homicides without gang truce

vi

We also found that the gang truce did not result in a homogenous decline in violence across

municipalities About 61 percent of municipalities experienced a decline in homicides but the

decline in violence varied substantially between municipalities We studied this issue further by

examining the impact of the initiative ldquoFree Violence Municipalitiesrdquo and it was found that the

behavior of violence in those municipalities was not significantly linked to the initiative but

rather to the general dynamic of the truce process nationwide Additionally we parsed out the

relative influence of the number of MS13 and 18th Street gang members on the street and in

prison from each municipality Our analyses indicated that following the truce the number of

MS13 and 18th Street gang members on the street in a municipality was not significantly

related to a decline in homicide but the number of imprisoned MS13 and 18th Street gang

members from the municipality was In particular the number of imprisoned MS13 gang

members from a municipality was associated with a significant decline in homicides in that

municipality following the gang truce and the number of imprisoned 18th Street members

from a municipality was associated with a significant increase in homicides in that

municipality following the truce These findings lend support to the idea that MS13 is more

organized than the typical street gang and that imprisoned MS13 gang members exhibit strong

influence over their fellow gang members on the street Our findings also suggest however

that the gang truce had a boomerang effect in municipalities with high numbers of imprisoned

18th Street members implying that 18th Street might not have as much organizational capacity

to regulate violence on the streets as MS13 The truce provided incarcerated MS13 and 18th

Street gang leaders an opportunity to negotiate with high-ranking officials and influential

diplomats including representatives of the Organization of American States This may have

increased their legitimacy inside and outside of their gangs It appears that MS13 was able to

exert its span of control over the communities in which they had influence and they were able

to deliver on the terms of the gang truce negotiations In the case of 18th Street however

incarcerated gang members may not have had the same organizational capacity for

communicating and carrying out directives In fact a review of the gang truce indicated that

there was a conflict taking place between two factions within 18th Street Consequently the

organizational structure and culture of 18th Street might be more diffuse than that of MS13

vii

and its leadership structure might not be as strong because of the internal fractures within the

gang This might further explain why homicides increased in 18th Street communities The

internal fractures within the 18th Street gang may have resulted in intra-gang violent conflict

that was largely contained within 18th Street controlled territories

Jamaica In Jamaica at first glance our impact findings appeared to show that the gang truce

might be an effective mechanism for reducing violence Bivariate analyses showed a significant

decline in homicides immediately after the truce was implemented This explains the work

previously published by policymakers researchers and news reporters Upon closer

examination of the data however comparing change in the target area to other areas in

Jamaica and accounting for temporal trends we found that the decline in homicide was part of

a larger nationwide decline in violence and that the gang truce was not responsible for the

decline The only significant effect that we uncovered was the possibility that homicides were

displaced outside the target area for a brief period of time but then returned to previous

levels

Honduras Our impact findings from our analysis of data from Honduras tell a very similar story

as Jamaica Initial analysis showed that the number of homicides on average declined across

municipalities following the gang truce Specifically the mean number of homicides declined by

12 per 100000 population from an average of 687 per 100000 population in each

municipality before the truce to an average of 566 thereafter However after we examined the

effect of the truce through time series analysis and included a variable (month) to control for

the temporal trends in the data the impact of the truce we observed in our bivariate analysis

was no longer significant Our findings as in Jamaica suggest that the decline in homicides was

less a consequence of the gang truce than of a broader short to medium term trend And

importantly in both Jamaica and Honduras the respective governments failed to deliver on

gangsrsquo demands for large-scale social and employment programs

Conclusions and Policy Implications

Over the last several years there have been a number of naturally occurring experiments

involving gang truces in a variety of nations and regions Findings from evaluations of gang

viii

truces are mixed In El Salvador the gang truce could be characterized as an effective short

term strategy to reducing homicides It is worth mentioning that despite homicides rates are

above truce levels they continue slightly below pre-truce levels In Jamaica and Honduras the

gang truce had no short or medium-term impact on overall violence In Los Angeles and

Trinidad y Tobago there was evidence that violence decreased for at least ninety days but then

increased substantially beyond those rates observed prior to the gang truce Thus the

effectiveness of the truce in El Salvador appears to be isolated and must be evaluated within

the context of other truces that have failed to reduce homicide violence Policy makers must

evaluate whether the conditions that allowed short term effectiveness of the gang truce in El

Salvador (such as the ability to promise and deliver immediate results) exist in other violent

areas before evaluating whether a truce strategy might be appropriate And they should be

heavily cautioned that the potential for long term negative consequences might outweigh the

potential for short term benefits

Indeed it is important to note that a number of scholars have noted that gang truces are likely

to result in a boomerang effect with gang violence increasing over the long run because of

enhanced cohesion within the gang (Klein 1995) Maguire (2013) notes that when government

officials negotiate a truce with gangs they might ldquoinadvertently be acknowledging gangs as

legitimate social entitiesrdquo (p 11) This in itself might increase cohesion among gangs which has

been found to be associated with increased levels of criminality (Decker et al 2008 Klein 1971

Maguire 2013) Hence it is important to consider the fact that gangs are illegal groups in El

Salvador and it should be cautious when carrying out dialog or negotiation processes with

them Further research is needed to examine how gang truces might impact group cohesion

and if it does whether the cohesion created could be effectively directed toward more

productive non-violent endeavors Gang truces convey the well-intentioned image that violence

has been addressed and policymakers are doing something about the problem but unless the

truce is implemented in a manner and under conditions where immediately achievable results

can be promised delivered and measured there remains a significant chance that the truce will

fail or worse yet backfire Thus it is imperative that any type of concession made by

ix

Governments to gangs within a truce framework should be transparent so that all sectors of

society have certainty that every action is being done within the existing rule of law2

In the case of El Salvador the truce arises from to the absence of effective public policies and

practices for violence control and prevention The truce was planned as a strategy to reduce

gang-related homicides During the process different organizations got involved including

religious (facilitating and protecting human rights) non-government (managing and facilitating

dialog and negotiation processes) international (providing funding for insertion programs) and

government (facilitating and providing certain conditions for dialog and negotiation)

organizations Some of the concessions that the Government provided in order to achieve a

reduction of homicides were within the law but others generated confusion and they seemed

to be close to the legal or socially acceptable limits This fact along with the poor transparency

of authorities in the management of the practice with public media and public weakened the

process and postponed its continuity

The present study suggests that gang truces should only be used as a means of last resort

and then only under certain conditions Given the risks associated with a gang truce

communities with high levels or at least modest levels of formal social control should rely on

other more promising gang control strategies such as pulling levers (ie Boston Ceasefire)

community oriented policing and the Gang Resistance Education and Training (aka GREAT)

program Only when the state has limited or greatly reduced capacity for social control should a

truce be considered Concomitantly a gang truce should be considered as an alternative only

when a community is experiencing a substantial amount of gang violence Communities that

are experiencing minimal to modest amount of gang violence may risk more from the

establishment of a gang truce than they have to gain Additionally our findings suggest that a

gang truce might only be feasible when gangs are sufficiently well organized to be able to

regulate their membersrsquo behavior and cause their members to behave less violently In El

22 Pragmatically this issue is complicated On the one hand transparency is a foundational element within a democracy and is necessary to ensure proper oversight of the government On the other hand it might not be possible to implement a gang truce with too many actors having a voice Policymakers might consider creating a policy that allows such negotiations take place but requires particular actors (such as a judicial body) to be informed of the process to ensure transparency and adherence to the rule of law

x

Salvador there is evidence of the strong organizational structure of gang MS13 Among other

factors the magnitude of its membership the chain of command from its leaders in prison and

the discipline of its leaders in the streets seemed confirmed On the contrary the

organizational structure of gang Barrio 18 ndash divided in two factions fighting over the leadership

ndash showed to be a less stable counterpart within the truce

Finally dialog and negotiations processes with or between gangs must have the capacity to

promise and deliver immediate benefits to the gangs that gang members can see or experience

in order to secure their continued participation in the truce as well as the capacity to monitor

and respond to truce violations Most importantly any effort aimed at reducing violence is

important and should be examined and assessed but it must have a transparent foundation

especially when it affects population rights as a whole

General recommendations

Gang truces are conjunctural strategies States who suffer from gang-related violence must

establish permanent public policies for crime control and prevention A government that

considers implementing a gang truce should be aware that it cannot become the center

strategy of its public policy for citizen safety

Gang truces should only be used as a means of last resort and then only under certain

conditions Stakeholders must determine whether a process of dialog or negotiation with gangs

is legal ethical and feasible

Stakeholders must anticipate demands that are likely to arise and their response options Some

demands may be easily met such as improved prison conditions Others are much more

difficult and amorphous such as community development through more integrated violence

prevention programs (such as those implemented by SolucionES in El Salvador) local economic

development programs or economic reinsertion of ex-gang members

Stakeholders should incorporate immediately achievable and demonstrable deliverables Long-

term goals and promises are unlikely to create the trust needed to sustain a gang truce

Stakeholders must first determine the position in which they are negotiating the incentives that

are possible to deliver and the boundaries and limits they face Gangs are mostly likely to trust

xi

representatives from NGOs community-based organizations and members of the faith based

community as brokers because they are considered more reliably neutral advocates for peace

They need to understand the capacity of the government to deliver promises in a timely

manner

Governments have to make a choice about the visibility and transparency of its participation

This decision needs to be made in the context of the national and local laws the publicrsquos

expectations of transparency and patterns of practices of the past

Governments must be strategic in their support for truce initiatives Some donor funded

programs run by the government prohibit gang member participation and if the government

does not receive approval from the donor it may risk the donor withdrawing its sponsorship of

the program

Governments must ensure an inter-institutional coordination for the management of truces to

avoid the responsibility to be of a single government institution It is necessary to generate or

collect reliable and pertinent data that can be used to analyze and assess the process

It is necessary to implement an effective monitoring system of the truce process similar to that

used in the full report as it can help parties understand what is working and who is delivering

on their promises More specifically the monitoring and truce management system should be

able to identify truce violations and be prepared to respond through the use of legal and

effective practices if stakeholders do not comply

Finally it is necessary to develop evaluations of gang truces and monitoring programs and

support violence prevention activities local economic development activities and pilot

programs to support the reinsertion of ex-gang members into society Clearly national

governments municipal governments NGOs and community-based organizations need

increased capacity and resources to discourage the growth of gangs among at-risk youth It is

therefore increasingly important to create economic opportunities for gang members willing to

leave the gangs and find other legal employment Developing and sustaining those

opportunities in nations with high incidences of poverty will require significant international

funding

xii

References

Cotton Paul 1992 ldquoViolence Decreases with Gang Trucerdquo JAMA 268(4) 443-44

Decker Scott H 2003 Policing Gangs and Youth Violence Belmont CA Wadsworth

Decker Scott H and W Reed 2002 Responding to Gangs Evaluation and Research

Washington DC National Institute of Justice

Decker Scott H Charles M Katz and Vincent J Webb 2008 ldquoUnderstanding the Black Box of

Gang Organization Implications for Involvement in Violent Crime Drug Sales and

Violent Victimizationrdquo Crime and Delinquency 54 153-72

Esbensen Finn-Aage and D W Osgood 1997 ldquoNational Evaluation of GREATrdquo US

Department of Justice Office of Justice Programs National Institute of Justice

Katz Charles M and Webb Vincent J 2006 Policing Gangs in America New York Cambridge

University Press

Klein Malcolm W 1995 The American Street Gang New York Oxford University Press

mdash 1971 Street Gangs and Street Workers Englewood Cliffs NJ Prentice-Hall

Maguire Edward Charles Katz and David Wilson 2013 ldquoThe Effects of a Gang Truce on Gang

Violencerdquo Unpublished paper Washington DC American University

Maguire Edward 2013 Research Theory and Speculation on Gang Truces Woodrow Wilson

International Center for Scholars in Washington DC

McCorkle R and T Miethe 2002 Panic The Social Construction of the Street Gang Problem

Upper Saddle River NJ Prentice-Hall

Ordog Gary J W Shoemaker Jonathan Wasserberger and Michael Bishop 1995 ldquoGunshot

Wounds Seen at a County Hospital Before and After a Riot and Gang Truce Part Twordquo

Journal of Trauma-Injury Infection amp Critical Care 38(3) 417-19

Ordog Gary J Jonathan Wasserberger Julius Ibanez Michael Bishop Eduardo Velayos

Subramaniam Balasubramanium and William Shoemaker 1993 ldquoIncidence of Gunshot

Wounds at a County Hospital Following the Los Angeles Riot and a Gang Trucerdquo Journal

of Trauma 34 779-82

Whitehill J M Webster D W amp Vernick J S 2012 ldquoStreet conflict mediation to prevent

youth violence Conflict characteristics and outcomesrdquo Injury prevention injuryprev-

2012

1

Introduction

Given the increasingly devastating individual family and community effects of gang activity

over the past several decades an increasing body of literature has focused on gangs gang

members and gang activity A core theme running throughout this body of literature is that

gang members are significantly more likely to be the offenders and victims of violent crime than

non-gang members (Battin Hill Abbott Catalano and Hawkins 1998 Esbensen Winfree He

and Taylor 2001 Katz et al 2011 Huff 1998 Curry et al 2002 Miller and Brunson 2000

Miller and Decker 2001 Pyrooz et al 2012 Curry et al 2001) and disproportionately affect

neighborhood levels of crime and violence (Block 2000) These findings have been robust in

that they have been repeatedly found regardless of research methodology (ie use of official

data self-report data observational data) (eg Curry 2000 Decker 1996 Deschenes and

Esbensen 1999 Esbensen et al 2001 Gordon et al 2004 Katz Webb Schaefer 2000 Webb

Katz Decker 2006) or research setting (ie North America Europe Asia South America

Australia) (Klein and Maxson 2006) As a consequence it should not be surprising that

policymakers academics and community leaders have focused much of their attention on

developing responses to address community gang problems

Suppression strategies have been the favored public policy response to gangs since the 1980s

(Venkatesh 1999 Spergel et al 1995) Suppression strategies typically rely on focusing

criminal justice resources on gang members through such practices as targeted and enhanced

police patrols intelligence databases vertical prosecution and enhanced sentences for those

convicted (Katz and Webb 2006) Suppression strategies are based on deterrence theory and

are founded on the principal that swift certain and severe penalties for those involved in gangs

and gang activities will necessarily result in fewer individuals joining gangs and will deter people

from engaging in gang crime (Klein 1995) By the early-to-mid 1990s as gang problems

continued to proliferate and policymakers sought alternative gang control strategies gang

prevention programming flourished (Papachristos 2013) Gang prevention programs were

aimed at the general youth population or focused on at-risk youth or neighborhoods Gang

prevention programming was based on the premise that by reducing risk factors and increasing

2

protective factors prevention specialists could inoculate youth from gang membership

(Esbensen 2000) Prevention programming has most frequently come in the form of school-

based prevention programming such as the Gang Resistance Education and Training (GREAT)

program where students are exposed to a curriculum designed to reduce both participation in

gangs and gang crime as well as the SolucionES3 program which engages youth and their

families through integrated community-based and school based programs

While a burgeoning body of literature has developed describing the assumptions issues and

effectiveness of suppression (eg Decker 2003 Decker and Reed 2002 Katz and Webb 2006

McCorkle and Miethe 2002) and prevention strategies (Esbensen and Osgood 1997) much less

attention has been given to examining gang intervention programming This might largely be

the consequence of the absence of such programming over the past 30 to 40 years By the

1980s policymakers no longer believed that social intervention approaches were an effective

strategy to control gangs and gang problems Although gang intervention strategies took many

forms they were based on the assumption that gang membership is the by-product of a socially

deprived community and that the values and norms of gang youth can be influenced and

directed toward those of mainstream society As such intervention programs often focused on

diverting youth from gangs or sought to minimize the consequences of gangs and gang activity

Gang intervention strategies include crisis intervention dispute resolution street-level

counseling and youth outreach (Spergel 1995) However many policymakers community

organizations and academics have argued that such approaches not only did not reduce gang

activity but may have lead to increased group cohesiveness which in turn may have lead to

increased delinquency (Klein 1971 Spergel 1995) More recent research has yielded similar

results For example a number of studies examining replications of Chicago CeaseFireCure

3 SolucionES is a Global Development Alliance in which $20 million is provided by USAID and SolucionES Alliance

members will contribute an additional $22 million dollars during a 66-month period starting July 2012 SolucionES is being implemented by an Alliance of five leading Salvadoran non-profit organizations who have come together to prevent crime and violence The Alliance members are Fundacioacuten Nacional para el Desarrollo (FUNDE) Fundacioacuten Salvadorentildea para la Salud y el Desarrollo Humano (FUSAL) Fundacioacuten Crisaacutelida (known locally as Glasswing) Fundacioacuten Salvadorentildea para el Desarrollo Econoacutemico y Social (FUSADES) and Fundacioacuten Empresarial para el Desarrollo Educativo (FEPADE) Together these five organizations have notable expertise in education health community development economic development research and youth leadership they are bringing their combined synergy and strengths to prevent crime and violence in El Salvador More information on social violence prevention programming is presented in Annex E including a summary of the violence prevention initiatives being implemented by SolucionES

3

Violence which relies heavily on crisis intervention dispute resolution street-level counseling

and youth outreach have found these strategies to either be ineffective or worse increase

levels of violence (Fox et al 2014)

One type of gang intervention gang truces however has received little attention in the

literature This is somewhat surprising given its use and public claims of effectiveness The

purpose of this report is to systematically examine the effectiveness of gang truces Gang truces

have largely been a ldquoblack boxrdquo that is little is known about the frequency of their use

conditions that give rise to them the role of third parties in brokering gang truces the

transformative effects of truces and the effectiveness of gang truces In this report we review

prior research on gang truces and present evidence on the effectiveness of gang truces that

have been implemented in El Salvador Jamaica and Honduras

Theoretical and Policy Rationales for Gang Truces

Much of the concern about gangs over the past twenty years has been their close association

with violence Qualitative and quantitative research has repeatedly found that gangs and gang

members are involved in high levels of serious violent offending Decker (1996) attributes the

gang-violence relationship in part to the collective and normative structure of gangs which

supports and encourages the use of violence both preemptively and in retaliation He further

explains that gang membership encourages the use of violence in retaliation against threats

and attacks whether actual or perceived which results in patterns of inter-gang conflict

characterized by escalating violence Related Cooney (1998) points out that gangs are not all

that different from ldquowarrior societiesrdquo He argues that while gangs are violent the violence

that they engage in is different than that engaged in by non-gang members in that it commonly

takes on a feud-like dimension A perceived slight violation of turf or other disrespectful

action might invoke a shooting which in turn evokes a retaliatory shooting which in turn

results in another retaliatory shooting creating a self-perpetuating cycle of violence

Early research and theoretical work examining gangs and gang conflict suggested that much

violent gang behavior was the function of status management (Thrasher 1927 Whyte 1943

Short and Strodtbeck 1965) Borrowing from the sociology and social psychology literature on

4

impression management theory gang scholars hypothesized that youth place significant

emphasis on image managementmdashseeking to impress their peers and limiting the potential to

embarrass themselves in front of others (Hughes 2005) These early gang theorists postulated

that violence is an instrument used by gangs and gang members to achieve manage and

protect status They contend that gang members often make decisions to become involved in a

violent conflict based on rational processes that weigh the immediate loss or gain of status

within the gang against the relatively small probability of being formally sanctioned by officials

within the criminal justice system (Hughes 2005) A number of researchers have examined the

relationship between status considerations and gang violence and have found the association

to be particularly robust regardless of gender (Campbell 1991 Giordano 1978) ethnicity

(Anderson 1990 1998 Vigil 1988) and location (Jankowski 1991 Spergel 1995) Hughes

(2005) notes that the centrality of status consciousness by gang youth may account for the

existence of the facilitative gang effect that gang scholars have repeatedly observed (ie Gangs

increase the amount of delinquency individuals are involved in beyond that of a group of

delinquent peers (Gordon et al 2004 Thornberry et al 2003)

Another micro-social factor associated with gang violence is group cohesiveness While our

understanding of the relationship between gang cohesiveness and violence is underdeveloped

some scholars have reported a strong relationship between the two In particular Klein and his

associates (1971 1967) and Lucore (1975) have reported that increases in gang cohesiveness

lengthens periods of gang membership and increases members participation in gang crime and

violence Cooney points out that there are strong relational ties between gang members that

necessarily result in increased cohesiveness among members For example gangs are

comprised of neighborhood youth who share common cultural and economic experiences and

have often known one another for lengthy periods of time He also points out that gangs are

groups that have strong self-proclaimed and formalized identities (eg colors symbols names

monikers) and have at least some organizational structure Their sense of group is maintained

by their common understanding of their members and friends and their attachment to their

territory (or neighborhood) Decker (1996) notes that the relatively high level of group

5

cohesiveness exhibited by gangs facilitates both collective behavior and liability (For exception

see Hughes 2013)

Accordingly both micro-social factors (ie status management gang cohesiveness) serve to

augment levels of gang violence and make it difficult for third parties to intervene Violence

within the context of gangs serves as a form of informal social control Gangs and gang

members cannot seek assistance from legitimate institutions of social control to solve conflicts

because they would risk losing status (Anderson 1999) Likewise the collective nature of the

gang not only increases potential offenders and targets of violence but also facilitates at the

group level the need for retributive justice on the behalf of injured members Moreover gang

members desire to impress others with their commitment to the group and use violence to

demonstrate their commitment to their group and to increase their status within the group All

of this results in an increased cycle of gang conflict and violence4

Over the past several decades residents and policymakers have expected the police to address

violent gang conflicts However there are several limitations to the police response to gangs

First as noted above many of the same factors that are associated with violent gang conflict

also limit the effectiveness of the police to have an impact on violent gang conflict Gang

members are not going to contact the police to resolve a conflict because it could result in loss

of status and expose them to the police discovering illegal activities that they are involved in

(Katz 2001 Katz and Webb 2006) Related citizens in neighborhoods with gang problems are

also reluctant to call the police out of fear of gang reprisals (Katz 1997 Webb and Katz 2003)

or because they have a poor perception of the police themselves (Katz Choate and Webb

2002) Third the police response to gangs in most communities is a reactive response to a

specific incident after it has occurred rather than a proactive response to intervene in ongoing

disputes between gangs (Katz and Webb 2003 2006) Most police agencies simply do not have

4 There is anecdotal evidence for these hypotheses Hughes and Short (2005) through field observations and interviews with members of 20 Chicago gangs examined the micro-social context of violent-related incidents Specifically they were interested in the specific social conditions that precipitated violence Their analysis revealed that gang disputes associated with retaliation escalated into violence significantly more often (ie 55 of the time) than disputes associated with normative order violations or an identity attacks which only resulted in violence roughly 33 percent of time

6

the intelligence networks required to intervene in gang conflicts until after they have risen to

relatively high levels (Katz 2003) Fourth police suppression strategies have been linked to

increasing gang cohesiveness and possible increases in gang crime (Klein 1995)

Some policymakers and community activists have proffered that an alternative to reliance on

formal mechanisms of social control such as the police is the gang truce (Spergel 1995) The

goal of a gang truce is to reduce or even eliminate violent conflict between those gangs that are

warring with one another As such compared with other strategies that often seek to reduce

general levels of crime a gang truce has the very specific goal of reducing violence between

two or more gangs that are in conflict with one another Unfortunately the literature provides

little guidance on the theoretical assumptions of why a gang truce should reduce inter-gang

violence Henderson and Leng (1999) hypothesize that at the root of gang truces is the notion

that they involve the renegotiation of existing norms within and between gangs The authors

claim that as a violent dispute escalates between gangs leaders and members are placed in the

situation of appearing weak to both members of their own gang and to members of the rival

gang if they do not respond with the appropriate amount of force or if they were to suggest a

peaceful resolution to the dispute Accordingly Henderson and Leng argue that as the cycle of

violence escalates between two or more gangs behavioral norms shift toward the increased

valuation of violence to resolve the conflict because it is the only option readily available to

them A gang truce on the other hand which is often mediated by a third-party is believed to

break the cycle of violence by providing the gangs involved in the dispute with a cooling-off

period (Spergel 1995) In the interim period new norms of expected behavior within and

between gangs are established In other words the cooling-off period is believed to recalibrate

norms of behavior that are more consistent with the security interests of the gang and its

members (Henderson and Leng 1999)

For many of the reasons stated above a number of communities have participated in gang

truces Gang truces have been observed in the United States Central America and the

Caribbean and gang leaders in other nations are considering its implantation (Fahah 2012)

Unfortunately little is known about gang truces We do not know when they come into

consideration how they are implemented whether they decrease increase or even have a

7

significant impact on violence In the below section we systematically review the existing body

of literature on gang truces Appendix A describes the methodology we employed to conduct

the systematic review

Prior Research on Gang Truces

Little research to date has examined the effectiveness of gang truces and of the research that

has been conducted has been restricted to the field of public health The most celebrated

truce to date was Philadelphiarsquos House of UMOJA 1974 gang truce In preparation of the truce

gang leaders in prison were called upon for their support along with key community leaders

from churches businesses schools and the police When UMOJA called for the gang summit it

was reported that 500 gang members and 75 of Philadelphia gangs attended By the

conclusion of the gang summit a 60-day truce was announced which reportedly resulted in no

gang members being killed over the 60-day period However no systematic evaluation of the

gang truce was conducted (Woodson 1981)

The first attempt to evaluate a gang truce to our knowledge was conducted by Cotton (1992)

who examined the results of a gang truce in South Central Los Angeles between the Crips and

Bloods Data provided by the police department indicated that over the 6-week period when

the truce took place drive-by shootings decreased by 48 compared to the same 6-week

period the prior year decreasing from 162 to 85 Likewise gang-related homicides dropped by

62 from 26 to 105

Ordog et al (1993 1995) examined the effects of a gang truce in Los Angeles using emergency

room admissions data Specifically the authors examined changes in the daily and monthly

number of gun shoot wound (GSW) emergency room admissions before during and after a

gang truce The catchment area for the emergency room was 100 square miles The authors

noted that while they were able to clearly identify the date the gang truce began because of

the media coverage that it received there was no specified date that the truce ended and as a

5 The authors did not report whether the drop in gang-related homicides was citywide or in the South Central Los Angeles neighborhood where the gang truce took place

8

consequence after 12 weeks the authors regarded it as called off for evaluation purposes

because it was no longer being discussed in the media and gun shot wounds began to increase

Student t-tests were used to examine changes before during and after the gang truce Ordog

et al (1995) reported that there were approximately 7 GSW admissions per day the 12-months

preceding the truce compared to 45 GSW admissions per day during the gang truce and 126

GSW admissions per day in the 11-months following the gang truce The authors concluded

that their analysis ldquoclearly showed that the institution of a gang truce had reduced the number

of GSW victims seen in an hellip inner city Level I trauma centerrdquo (p 419)6

However it is important to point out that while the gang truce in Los Angeles did decrease

homicides by about 35 percent for the first three months it then doubled in months four

through eleven compared with the pre-truce period - a pattern that the authors did not

discuss Similar findings were reported in Trinidad and Tobago (Maguire Katz and Wilson

2013) where it was determined that homicides declined for a brief period of time (again for

about three months) but then increased substantially over the long term (12 months) These

results suggest that gang truces may produce short-term benefits yet result in long-term

adverse consequences

While the research examining gang truces shows their potential promise and their potential for

greater harm we believe that the findings should be viewed with caution for two reasons

First some of the implicit theoretical assumptions that gang truces are built on may not be

accurate Prior research on gangs suggests that they have limited organizational structure

(Decker Katz and Webb 2008 Decker Bynum Weisel 1998) and have few formal

mechanisms to influence member behavior For example most gangs do not have formal

leaders do not require members to pay dues and members do not make contributions to the

gang for the purpose of developing the gangs infrastructure (ie guns housing etc) A number

of academics have pointed out that ldquosophisticated gang organizations is still largely a product of

the self- or organizational-interested musings of gang leaders certain police officials academic

6 Similar results were presented by Ordog et al 1993 using the same data but analyzing it over a shorter period of time and aggregating the data by month instead of by day

9

researchers media reporters based on very limited hard datardquo (Spergel 1995 79-80)

Therefore even if a truce was successfully negotiated between members of gangs in conflict

much of the academic research suggests that gang leaders do not have enough control over

members to enforce a truce Related prior research on gangs and gang members suggests that

they are primarily comprised of young people with few ldquodiplomaticrdquo skills (Henderson and Leng

1999) The very same organizational and normative features of gangs that result in gang

violence (ie often bastions of young minority male street youth) are those same features that

most likely limit the likelihood of a gang having the capacity to abide by a gang truce In sum

implicit in the theoretical assumptions of a gang truce is that gangs have the organizational and

cultural capacity to create and maintain (at least for a short while) a truce which is inconsistent

with much prior research on the organizational characteristics of gangs

The second reason we believe that prior evaluations of gang truces should be viewed with

skepticism is that they have relied on relatively weak evaluation designs For example prior

studies examining gang truces have not incorporated the use of comparison areas or control

groups The causes of reductions in gang violence found in previous evaluations might be

many For example the Los Angeles riots took place just before the gang truce that Ordog et al

evaluated Zinzun (1997) reported that gang culture and violence changed briefly but abruptly

following the riots because gangs and gang members in part redirected their anger and focus

toward the police As such the decline in GSW admissions may have been the consequence of

an overall city-wide decline in gang violence in the wake of the LA riots Related prior research

has relied on fairly broadly defined outcome measures such as GSW admissions and general

levels of drive-by shootings Such outcome measures lack specificity in terms of attributing

gang violence to the specific gangs involved in the gang truce Evaluation designs used in

previous studies have also lacked specificity in terms of the catchment area where violent

activity was labeled as gang related For example Ordog et alrsquos (1993 1995) outcome measure

included all emergency room admissions for GSW However the emergency room received

patients from a 100 mile square area surrounding the hospital an area that was most likely

much larger than the gangsrsquo territories involved in the truce

10

Additionally prior research examining gang truces has not examined the processes involved in

the creation of the gang truce No context has been provided in terms of the factors that lead

to the gang truces whether the gangs were pushed or pulled into truces whether outside

parties helped mediate the truces or whether on-going mediation was required to maintain the

truce In other words we still do not know the processes related to the formation of a

successful gang truce Instead prior research and evaluation has treated the gang truce much

like a black box where it is described in very general terms but its details are not revealed

Likewise prior evaluations have not examined the effects of a gang truce beyond its impact on

gun violence A number of prior researchers have argued that mediating such activities serves

to legitimize gang leaders increases gang identity for members and results in greater group

cohesion (Haskell and Yablonsky 1982 Klein 1995) As such we do not know if there are any

unintended consequences resulting from gang truces due to the methodologies chosen for

prior research and evaluation Finally prior research on gang truces has not relied on more

sophisticated statistical analysis that has the capacity to not only discern overall trends but also

cyclic patterns outliers and turning points

THE PRESENT STUDY

The present study seeks to understand the effectiveness of the gang truce when negotiations

occur with or between gangs and with government institutions or civil society organizations

We have briefly reviewed the theoretical and policy rationales for gang truces and the

literature and prior research on this form of gang intervention Next we present three case

studies one from El Salvador another from Jamaica and another from Honduras Each offers

evidence about their impact on violence and lessons learned that are unique to their particular

circumstances In conclusion the case studies are considered together and we present their

collective policy implications

11

Case Study A Gang Trucemdash The Salvadorian Experience

Over the past two years the Salvadoran gang truce brokered by an array of local policymakers

and international development organizations has attracted national and international

attention Violence in El Salvador has been at an all time high with a rate of 692 homicides per

100000 population in 2011 (UNODC 2014) making El Salvador one of the most violent nations

in the world (Parkinson 2014) Since 1992 government and non-government actors have been

responding to the rising tide of gang violence using traditional suppression strategies such as

increased policing legislative changes and more severe prison sentences These traditional

mechanisms of formal social control however were proving ineffective if not counter-

productive (Perez 2003)

In response to the inadequacy of traditional strategies stakeholders altered their course in an

effort to radically reduce gang violence in the nation Members of the Funes administration led

a group of negotiators comprised of the Catholic Church a former congressman and the

Organization of the American States (OAS) to help frame the conditions for a possible truce

between the MS13 and 18th Street gangs (Umana de Leon and Tager 2014) In March 2012 a

truce was reached The goal of the gang truce was to reduce violence specifically gang-involved

homicides Included in the terms in exchange for the gangs acting to reduce homicides certain

incarcerated gang members were to be transferred to lower security prisons to receive

sentence reductions and special visitation privileges and to be permitted more communication

with those outside of the prison for the purpose of conducting crisis interventions to mitigate

the violence (Salanegra 2012)7 The gang leaders also agreed to no longer recruit children into

their gangs reduce violence against women give up a small number of guns and continue to

participate in negotiations (Seelke 2014 11-12)

7 It is important to note that there were other discussions that took place during the mediations that included as a local publication noted ldquoProhibition Act Gang send the army to barracks end the police operations in controlled areas by gangs repeal of the figure of the witness criteriado (with criteria of opportunity or ldquowitness under a plea agreementrdquo) and a number of improvements in the quality of life of the inmates and its familiesrdquo Source httpwwwsalanegraelfaronetes201209cronicas9612

12

The present case study seeks to understand the role of negotiations with or between gangs and

government institutions or civil society organizations and their impact on violence in El

Salvador The specific objectives of this case study were to (1) Identify and document the

negotiation processes with or between gangs This includes but will not be limited to

identifying the actors involved in the negotiations the goal(s) of the negotiations and

strategies employed to carry out negotiations (2) Identify the impact of negotiations with

andor between gangs on violence and other forms of criminality and (3) Present conclusions

about the strengths and weaknesses of negotiations and their potential for positively or

negatively reducing crime in Salvadoran society

The Salvadorian Gang Truce

While the origins of the Salvadorian gang truce are somewhat unclear some accounts suggest

that in the fall 2011 the Minister of Security (then David Munguia Payes) mentioned the idea of

starting a dialogue between the gangs to Raul Mijango (a former congressman) Shortly

thereafter Mijango spoke to Monseigneur Fabio Colindres the military chaplin about the idea

They started to communicate with gang leaders in January 2012 From the beginning the

Minister of Security stated that he would not personally communicate with gang leaders and

noted that the official position of the government was not to negotiate with offenders But he

did allude to the fact that the gang truce was part of a new strategy to address the nationrsquos

gang problem (Archibold 2012) The Salvadorian gang truce was multi-dimensional involving a

varying number of actors communication styles and tactics In the section below we discuss

the parties involved in the truce the negotiation process and the strategies used by the gangs

and government in furtherance of the truce

Parties involved in the truce There were a series of persons groups and other entities who

directly or indirectly participated in the negotiation process whether they were negotiators or

collaborators From the beginning negotiations between the MS13 and 18th Street gangs were

considered confidential and were held in strict confidence between the gangs and the

negotiators Beginning in early 2012 the first two mediators to serve were a former

congressman (Rauacutel Mijango) and a priest (Fabio Colindres) These individuals were able to gain

13

the trust of the gangs and also had open communications with the government and

community

As the process moved forward however church leadership changed its initial position and

withdrew its support of the truce This left the priest to participate as a private individual no

longer serving as a representative of the church The mediators had some early success

however as the mediators changed the lines of communication became less clear and less

reliable Still they played an important role in the process Finding individuals whom the gangs

would trust and who could speak for government and community institutions was challenging

Further as government ministers and presidents changed it became less clear whether or not

the government was participating in the dialogue between various parties much of the

confusion in negotiations was related to understanding the governmentrsquos official role in the

process

In the beginning the negotiating parties included the two mediators and MS13 and 18th Street

the two predominant gangs in El Salvador MS13 is the largest gang in the country with about

250 cliques throughout the nation For the most part its first- and second-generation leaders

were in prison There are municipalities in which only MS13 operates and therefore its

members never clashed with 18th Street adversaries MS13 members are largely involved in

extortion violence and intimidation of the public The MS13 leadership has been shown to

have greater control of its organization than most its members have been more disciplined

and despite its size the gang seems to have sufficient and effective communication

mechanisms between cliques and its leadership (ldquoranflardquo or ldquoruedardquo)

The 18 Street gang is somewhat smaller than MS13 Long before the truce process began

factions existed within 18th street which had become deeply involved in internal conflicts over

leadership structure Those differences had become more pronounced over time leading to the

establishment of two movements known as the ldquoSurentildeosrdquo (Southerners) and the

ldquoRevolucionariosrdquo (Revolutionaries) that participated in the peace process but under different

14

representation Despite their differences however for the negotiation process the two

fractions of 18th street established similar demands between themselves and with MS138

As noted above the governmentrsquos involvement in the truce process was never clearly

articulated and it varied with ministerial changes The Funes Administration specifically the

Ministry of Justice and Public Security (MJPS) had been engaged in the negotiation process

early on In 2013 however with the change in MJPS leadership this also changed Tensions

grew over the role of the government in the truce negotiations According to the mediators the

new Minister of Security wanted to dismantle the process and for this reason he blocked the

mediatorsrsquo entry to the prisons where they had been meeting with gang leaders The change in

government leadership brought changes in strategy As with all negotiations mutual trust was

the cornerstone When the personnel and their agendas changed so did the trust in the peace

process

Late in 2012 as negotiations began to focus on the relationship between the gangs and the

community other organizations were called upon to help facilitate this interaction A technical

committee was created to oversee progress towards the violence reduction goal this was

announced at the end of August 2012 The committee moved forward with support from the

Organization of American States (OAS) It coordinated actions intended to improve the living

conditions of incarcerated persons to reintegrate into society those who had served their time

in prison and to prevent violence and provide assistance to victims of violence The committee

operated as a mechanism for coordinating the negotiation process with technical support from

the United Nations Development Program (UNDP)‒El Salvador Interviews with experts

indicated that Fundacioacuten Humanitaria with support from several organizations like Interpeace

and others had also supported the operations of the mediators and served as an important

point of reference for the gangs

In each territory where agreements were made government representatives from some

municipalities were invited to participate in the process as key implementation stakeholders

Between December 2012 and January 2013 the municipalities of Ilopango Santa Tecla

8 httpsalanegraelfaronetes201110cronicas5645

15

Sonsonate and Quezaltepeque became among the first to become part of what later became

known as the Sacred Municipalities (aka violence free municipalities) By November 2013

several other municipalities had become part of this group La Libertad Apopa San Vicente

Puerto El Triunfo Nueva Concepcioacuten Ciudad Delgado and Zacatecoluca

The gangs and the government offered several incentives to one another The gangs offered a

number of actions to reduce the incidence of crime (ie cease of hostilities zero homicides

stop extortions robberies and recruitment of children and youths into gangs peace at the

schools) and the government offered to guide public investments in social policies prioritizing

participating municipalities

The truce process was the product of a convergence of vested interests from different sectors

Several earlier attempts at negotiations had failed likely because of a lack of perceived

authenticity on the part of the actors All parties involved in this truce process came to the

table with well-defined group interests and concerns All sides were experiencing fatigue from

the long history of violence Years of fighting deaths and violence had led the mostly

incarcerated first generation those who were still living and who continued to serve as gang

leaders to seek new ways of exercising social power and influence in prison and the

community

The negotiation process In El Salvador the negotiation process between gangs formally began

on March 8 2012 by a decision of the Government to favor the transfer of gang leaders

imprisoned out of the countryrsquos only maximum security prison located in the municipality of

Zacatecoluca That decision was made by the Minister of Justice and Public Security as part of

an agreement between a team of mediators who served as advisors to the Ministry and gang

leaders primarily from MS13 and 18th Street The negotiations were referred to as a ldquotruce

between gangsrdquo and were characterized by the mediators as ldquoa peace processrdquo or a ldquocease of

hostilities between gangsrdquo

Publicly the gang truce was known as an agreement between gangs and not between gangs

and the government because of fear of how those outside the negotiations might interpret the

governmentrsquos actions From the beginning therefore the government never fully

16

acknowledged its participation in the peace process (through representatives) with the gangs

and that reticence was a major impediment throughout the process This was largely because

of the difficulty in communication between the multiple parties which was done through

mediators and necessarily resulted in additional logistical complexity throughout the

negotiations Regardless it is important to recognize that the government played a decisive

and central role in the origin facilitation and promotion of the peace process

In terms of processes interviews with gang leaders indicated that they would communicate

their ldquodemandsrdquo to the mediators who in turn would communicate the gang leaderrsquos demands

to the government The same process was used for the government to communicate with the

gang leaders After an agreement was reached the parties would make proclamations that

emphasized the agreed upon terms of the negotiations These often took the form of public

statements made through local newspapers For example there were more than twenty public

statements made by the gangs where they clearly articulated that their intention was to

reduce the harm that their acts of violence were causing themselves and the communities (For

a detailed discussion of the public statements see Appendix B) In these public statements the

gangs acknowledged the need for their groups to change Both the mediators and the gangs

made clear that the gangs expected the government to respect and respond to their demands

given the good-faith actions that the gangs had already taken

The government on the other hand did not make any formal public statements about their

end of the agreements reached through the negotiations From the outside it appeared that the

government did not develop a formal or consistent policy regarding the truce and instead

assumed that they could achieve a reduction in homicides through the negotiation of increased

prison privileges for gang leaders At the close of the Funes administration the peace process

and negotiations initiated within its framework had reached a plateau There they remained

waiting for the new government to step in and take action as President-elect Salvador Saacutenchez

had announced his support for the truce during the electoral process Sanchez however

withdrew that support when he assumed the presidency in June 2014 Indeed under the

recently elected Sanchez administration the government changed course and became less

17

willing to engage the gangs believing that the government should not negotiate with criminal

organizations

Indeed it should be noted that responses resulting in increased gang influence have been a

concern over the past several decades--from governmental neglect of the problem in the

1990s through the period of the mano dura and super mano dura of the early 1990s to

2000 and to the present gang truce Many have asserted this claim as truth (ie some

responses will increase the influence of gangs) especially in public discourse This study

however does not directly address this point which may or may not be valid9

Strategies employed during negotiations between the gangs and the government Over the

course of the negotiations between the gangs and the gangs and the government at least two

strategies were employed At the beginning of the negotiations strategies were implemented

similar to that of the historic Salvadorian peace accords Each of the parties with the support of

mediators reached agreements and achieved a resolution to their conflict This model

although useful overlooked an important difference between the current violence in El

Salvador and the civil war of the 1980s In a conflict of civilians with the State the legitimacy of

the ldquoadversariesrdquo arises by the need of the state to recover the rule of law It is certainly

possible to reach peace agreements that resolve conflict between gangs but in the current

case with regard to the government there was no legal or policy justification for executing the

truce process On the contrarymdashgovernment negotiations with a criminal group are relatively

rare in modern democratic societies The lack of a legal or policy framework to work from

limited the governmentrsquos ability to be transparent in its response to gangs and may have

undermined its legitimacy with the public

The second strategy employed by the parties was based on reciprocation and cooperation

between actors Early on it was determined by both sides that if one of the parties abandoned

the peace process or did not hold up its end of an agreement the other party would no longer

participate in the negotiation process (Axelrod 1986) In retrospect during the course of the

peace process the gangs implemented agreed-upon terms and positioned themselves as valid

9 We would like to thank one anonymous reviewer for bringing this issue to our attention

18

partners with the government able to negotiate for what is needed by the government in

exchange for what the gang needed10 For example in public statements the gangs insisted

that they had made a goodwill gesture when they declared a unilateral truce and stopped their

involvement in violence The government responded by transferring incarcerated gang leaders

from a maximum-security prison to ordinary criminal prisons11 Next the gangs offered to

disarm the government responded with increased flexibility on control measures at those

prisons This process continued until the demands from both sides grew in a direction that

challenged each sidersquos capacity and willingness to follow through

For instance mediators through the government asked the gangs to end the extortion of

businesses and individuals which are the primary means of subsistence for Salvadorian gangs

The gangs asked that the government reciprocate by eliminating the ldquowitness under a plea

agreementrdquo from the criminal procedural law which is one of the main weapon laws used for

sentencing gang members for complex crimes such as extortion Each of these requests were

more than the parties could carry out Exhibit 1 outlines the different ldquooffersrdquo and ldquodemandsrdquo

made by each party The offers and demands are divided into those that were believed to be

relatively simple and were ldquoofferedrdquo by the party on its own volition and those that were

believed to be more complex and were demanded by the other party

Exhibit 1 Simple demands and concessions made by gangs and the government

Gang concessions Government concessions

Reduce the homicides by more than 50 Transfer the leaders to prisons with lesser security

Hand over 500 firearms to the authorities Allow visits of the children of the gang members

Reduce violent actions at schools Allow night intimate visits

Stop killings at the prisons Allow entry of fast foods

Do not murder custodians police soldiers or their family members

Facilitate the entry into the prison of gang members let out of prison

Give opportunities to some gang members to withdraw

Suspend the searches by the armed forces on persons at the prisons

10 Source httpwwwsalanegraelfaronetes201209cronicas9612 11 This was useful to facilitate the coordination and the communication between the gang members in prisons and in the streets

19

Make it more flexible for new social plans by government to be established in the communities controlled by gangs

Improve the conditions and health assistance at the prisons (eg place tvrsquos to improve mental health of inmates)

Do not burn buses do not kill bus drivers or collecting agents in buses

Allow the Red Cross and journalists to enter into the prisons

The second agenda presented in Exhibit 2 shows the more complex demands that were

not offered by one side or the other but rather were demands placed on the other party These

demands were such that they required a higher level of authority to negotiate in order to

implement the demanded action These demands were considered critical for the peace

process to continue and the delayed responses on both sides stalled the progress of the truce

and led to the parties questioning the legitimacy of the other side all these demands have yet

to be attained The mediators recognized at the end of 2012 that some of the slow progress

was related to the difficulty of making the transition from offers that could be executed

relatively simply to the more complex demands made by each party One of the gang leaders

summarized the issue ldquoWe are not looking for television sets while all of our people continue

living like shit we are not going to try to do everything that is in our power to decrease as much

violence as possible for one television set There are things we are very clear about this [points

to a television set in the room] is a right that the law grantsrdquo

Exhibit 2 Complex demands and concessions made by gangs and the government

Demands by Gangs Demands by Government

Eliminate the figure of witness under a plea agreement from the Criminal Law

Stop homicides and extortions indefinitely

Create international commission to investigate the human rights violation cases of the gang members and their family members on the part of the State (PNC and FA)

Share the information regarding the whereabouts of the brunt of extortion money whether in country or abroad

Institutionalize external and professional surveillance regarding the behavior of the PNC as regards investigations and gang member arrests

Progressive dismantling nation-wide of the clique structures and turfs

Army should stay in its barracks and definitely suspend their participation in public security tasks

Permanently suspend the orders to murder State security and justice agents and their family members (police soldiers judges prosecutors)

20

Clearing of judges prosecutors and police involved in corruption cases against youths in conflict with the law

Permanently suspend the murders extortions and harm caused to public transportation resources and their workers

Maintained a sustained improvement on the conditions of the prisons

Permanently suspend the recruitment of children and youths and hostilities to educational centers

Guarantee working opportunities for the gang members and their family members through specific programs at the municipalities

Share the information about providers of drugs and illegal arms

Television sets were one of the concessions made by the government from the simple list The

gang leader recognized that their intentions in the negotiations extended beyond improving

prison conditions however the government did not have support nor established mechanisms

for carrying out tasks derived from the more demanding list Further as the government

transitioned to new presidential leadership many of the more simple concessions had already

been made and the new government would have to address more complex demands if

negotiations were to continue

The Salvadorian gang truce is remarkable for several reasons First a number of policymakers

and researchers have claimed that the truce saved a large number of lives and was perhaps the

most successful gang truce in the Western Hemisphere Second the Salvadorian gang truce is

somewhat unique in that it might have lasted substantially longer than any other successfully

negotiated gang truce Understanding the temporal impact of the truce is important to

understanding its future potential Third the Salvadorian gang truce is important because a

number of other counties have sought to replicate it For example following the perceived

(and perhaps real) success of the Salvadorian gang truce the nations of Honduras Belize and

Guatemala instituted or tried to institute a similar type of truce Understanding the impact of

the Salvadorian gang truce will further help us understand the potential impact of such a

process on violence in these other nations

Methods

We examined the impact of the gang truce by merging four separate data sets First data from

the 2007 El Salvador Population and Housing Census provided municipal level measures of

21

percent unemployed percent male aged 10-29 percent female headed households ethnic

composition in-migration income percent urban percent households rented and percent

who had a high school education These data were obtained directly from El Salvadors Ministry

of Economy Second municipal level crime data (ie homicide extortion rape theft robbery

and auto theftrobbery) were provided by the National Civil Police (PNC) by month and year for

the period between January 1 2010 and June 30 2014 Disappearance data was also provided

by the police aggregated by year and municipality Third police gang intelligence data was

provided by the National Civil Police (PNC) Specifically we received the number of police

recorded MS 13 and 18th Street gang members by municipality in 2011 the latest year for

which these data were made available Last we acquired 2011 prison gang intelligence data on

the number of imprisoned MS 13 and 18th Street members by municipality from the

Salvadorian National Bureau of Prisons All four data sources were linked together using a

unique municipality identification number and month Collectively they provide an

opportunity to examine the impact of the gang truce in El Salvador and understand several

competing explanations for any changes that might have occurred

The geographic unit of analysis for the present study is the municipality According to the 2007

El Salvador Population and Housing Census there were 14 departments divided into 262

municipalities (the equivalent of a county in the USA) in El Salvador Of the 262 municipalities

ten (38) of the municipalities were eliminated from the analysis because of missing data

Measures

Dependent variables The dependent variable in the present study is the monthly number of

homicides in each El Salvadoran municipality expressed as the number of homicides in

municipality i at time t Homicide data were collected from January 1 2010 through June

30 2014 for a total of 54 months This includes data for a period of 26 months prior to the gang

truce and 28 months following the implementation of the gang truce As shown in Exhibit 3

there were a total of 14148 homicides over the study period with each municipality averaging

371 homicides per month

yit

22

Exhibit 3 Descriptive statistics

n mean sd

Urban Opportunity Factor 252 111 95

male 10-29 years old 252 1960 171

female-headed household 252 3433 499

unemployed 252 1177 686

Racialethnic heterogeneity 252 17 13

of prisoners MS13 252 1579 3960

of prisoners 18th Street 252 1268 4096

of prisoners MS13_spatial weight 252 1579 3960

of prisoners 18th Street _spatial weight 252 1268 4096

MS13 on the street 252 4146 10646

18th Street on the street 252 2513 8622

Violence free municipality 252 04 20

Truce 14148 371 783

Monthly homicide rate per 1000000 14148 51 49

Independent variables Several independent variables were included in our model Our

measure for the intervention is a dummy variable in which the value of 0 is used to represent

the pre-intervention months and 1 represents the post-intervention months We also

included a dummy variable for each community that participated in the violence-free

municipality initiative (sanctuary municipalities) to address issues of external validity (eg

multiple treatment inference) In other words municipalities that participated in the

violence-free initiative might have experienced or participated in something that either

enhanced or detracted from the impact of the larger gang truce The initiative called by some

the second phase of the gang truce took place about 8 to 9 months after the original gang

truce was negotiated in March 2012 In participating sanctuary municipalities gang members

agreed to stop violence and crime in exchange for a reduction in police operations and night

raids12 The municipalities that participated in the initiative included Santa Tecla

Quezaltepeque La Libertad Ilopango Ciudad Delgado Apopa Sonsonate San Vicente

Zacateatecoluca Puerto El Triunfo and Nueva Concepcioacuten

12 See httpwwwcispesorgblogviolence-free-cities-inaugurated-as-second-phase-of-gang-truce

Tt

Si

23

We included several independent variables related to the presence of gangs by gang for each

municipality Specifically we included count variables on the number of MS13 and 18th Street

gang members who were on the streets and in prison by municipality As noted before the

gang truce was made between a relatively small number of imprisoned gang leaders from both

MS13 and both fractions of 18th Street who agreed to stop street-level gang violence on the

condition that they would be transferred to lower security prisons and granted special

privileges

We wanted to understand the impact of the truce in the context of gang presence

Municipalities with high numbers of MS13 and 18th Street members whether they were in

prison or on the streets should have experienced a greater reduction in homicides because of a

greater span of control over these communities Accordingly municipalities with low numbers

of MS13 and 18th Street members should have experienced a lesser reduction in homicides

because of a limited span of control over these communities As Exhibit 3 (above) shows on

average municipalities had about sixteen MS13 and thirteen 18th Street members in prison

and forty-one MS13 and twenty-five 18th Street members on the street We examined whether

gang members in El Salvador were randomly distributed and found that there was spatial

clustering in the number of MS13 and 18th Street members who were in prison In order to

control for the clustering we included the spatial lag (weight) of the MS13 and 18th Street

members in prison

Control variables A series of control variables known through prior research to be related to

violence in communities were added using the census data We began with municipal-level

measures of percent unemployed percent male aged 10-29 percent female-headed

households ethnic composition in-migration income percent urban percent households

rented and percent with a high school education Initial analysis found that five community

variables were highly correlated and loaded on the same factor As seen in Exhibit 4 these

included the percentage of persons who had moved there from another municipality (in-

migration) average income in the municipality percent of the population living in an urban

area percent of houses that are rented and percent of residents with at least a high school

education We labeled this factor Urban Opportunity

24

Exhibit 4 Factor loadings for urban opportunity

Variables Factor

loadings

In Migration Percent of population moved in from another municipality 0638

Income Average monthly income per household (colones) 0886

Urban Percent of population that is urban 0845

Rented Percent of households rented 0761

Education Percent of residents who have at least a high school education 0742

Percent of variance 6071

Eigenvalue 341

Extraction method Principal Axis Factoring

We also calculated a measure of ethnic-heterogeneity from relevant census data Ethnic

heterogeneity which varies from 0 to 1 was calculated by taking one minus the squared

proportions of the population in each ethnic group (White Mestizo other) As with the

percentages of male population that is 10-29 years old female-headed households and

unemployed persons ethnic heterogeneity has consistently been associated with violence in

general (Kubrin and Weitzer 2003) and with gang violence in particular (Katz Maguire and

Roncek 2002 Rosenfeld Bray and Egley 1999)

Analytic Strategy

National-level analysis Our analysis began with a simple t-test of the mean number of

homicides at the national level before and after the truce providing the most basic omnibus

test of an effect We then presented our time series model to estimate the effect net of

seasonality and temporal trends This analysis included a set of simple ARMA (autoregressive

and moving average) models with two lag periods and one period of a moving average Initial

analysis of the number of homicides by month indicated that partial temporal autocorrelations

existed for two lags The first model used only data prior to the truce This model included a

linear time trend variable and dichotomous variables for each month (except January) We used

this model to forecast the expected number of homicides for the truce period This series of

analyses was for illustrative purposes only as the number of data points used in the forecast

25

was too small to provide meaningful confidence intervals The second ARMA model employed

all data from January 2010 through June 2014 and included a dichotomous indicator for the

truce period The effect of this variable was the average change in the number of homicides

net of seasonality (months) and temporal trends

Municipality multilevel models The analysis at the municipal level presented four challenges to

a typical regression model First since our outcome has a highly skewed distribution with

varying levels of exposure (population) we employed a generalized linear model to capture the

correct distribution In this case we employed a negative binomial distribution rather than a

Poisson model due to the over-dispersion in our outcome created by analyzing monthly

municipal datamdashthat is there were several months and municipalities where no or very few

homicides occurred

The second challenge was our need to measure the between-municipality variation of pre-truce

homicide rates and the program effect A fixed-effects model would have been inappropriate

because it would not have allowed us to estimate these variance components Thus we

employed a multilevel random effects generalized linear model

The third challenge was temporal autocorrelation Analysis at the national level indicated

partial autocorrelations in the first and second lags Although a generalized model does not

allow for auto-correlated residuals we addressed this by entering in the model two lags of the

homicide rate

The fourth and final challenge to this estimation was spatial autocorrelation An examination of

the Moranrsquos I and autocorrelation coefficients (exhibit not shown) indicated a low level of

autocorrelation for each month but many months were still statistically significant Thus we

estimated spatial lags of the homicide rate and entered them into the model

Accordingly we analyzed the data using random effects negative binomial models regressing

the rate of homicide on various months on predictors with both temporal and spatial lags For

clarity we present the model using HLM (Bryk et al 1996) notation At the first level (time) we

estimated the log number of homicides using a negative binomial distribution with predictors

26

that included the truce period calendar month linear month trend a

one period lag of the homicide rate a two-period lag of the homicide rate a

spatial lag of the homicide rate and an over-dispersion parameter which has a

Gamma distribution of Thus the final level 1 model is

At the municipality level (level 2) the intercept is a function of the presence of the

Violence Free Municipality program the log number of MS prisoners the log

number of 18th Street prisoners the log of the spatial lag number of MS prisoners

the log of the spatial lag of the number of 18th Street prisoners the log of the

number of MS members the log of the number of 18th Street prisoners the

control variables detailed above and the random effect Thus the level 2 model for the

intercept is

Also at the municipality level (level 2) the effect of truce is a function of SAFE the

log number of MS prisoners the log number of 18th Street prisoners the log of

the spatial lag number of MS prisoners the log of the spatial lag of the number of 18th

Street prisoners the log of the number of MS members the log of the number

of 18th Street prisoners and the random effect Thus the level 2 model for the

truce effect is

TRUCEt mt TIMEt

HRt-1i HRt-2i

HRti vti

Gamma1

aa

aelig

egraveccedilouml

oslashdivide

mti = expb0 i + b1iTRUCEt + lmmt

m=1

11

aring + l12TIMEt

+l13HRt-1i + l14HRt-2i + l15HRti + vti

eacute

euml

ecircecircecirc

ugrave

ucirc

uacuteuacuteuacute

exp vti[ ] ~ Gamma1

aa

aelig

egraveccedilouml

oslashdivide

b0 i

SAFEi PMSi

P18thi

PMSi P18thi

MMSi M18thi

Ci u0 i

b0 i = g 00 +g 01SAFEi +g 02 ln PMSi +1( ) +g 03 ln P18thi +1( ) +g 04 ln PMSi +1( )

+g 05 ln P18thi +1( ) +g 06 ln MMSi +1( ) +g 07 ln M18thi +1( ) + pC

C

aring Ci + u0i

b1i SAFEi

PMSi P18thi

PMSi

P18thi MMSi

M18thi u1i

27

The key parameters of our analysis are in this expression The truce effect is dependent on

the log number of MS prisoners among other characteristics Thus the percent

reduction in homicides can be calculated from this model using the following formula

which we plot below (see Quantitative Findings below)

To answer questions about the variation in the effects of the gang negotiations we predicted

best linear unbiased predictions (BLUP) of the Truce random effect which estimates the

between-community variance of the immediate effect of the truce Since we assumed that the

interventions were not evenly spread throughout communities in this study the mixed model is

appropriate to address this issue We conducted this analysis for homicide using a Stata

generalized linear mixed model using full maximum likelihood and an identity covariance matrix

of random effects

Findings

Exhibits 5 and 6 present our findings related to the number of police recorded homicides in El

Salvador from January 2010 through June 2014 The trends in the data and the results of the t-

test show that in the 26 months prior to the gang truce there were on average about 354

homicides per month compared with about 218 homicides per month in the 28 months

following the gang truce Exhibit 7 shows that the gang truce resulted in 5501 fewer homicides

than otherwise would have occurred

b1i = g 10 +g 11SAFEi +g 12 ln PMSi +1( ) +g 13 ln P18thi +1( ) +g 14 ln PMSi +1( )

+g 15 ln P18thi +1( ) +g 16 ln MMSi +1( ) +g 17 ln M18thi +1( ) + u1i

b1i

ln PMSi +1( )

1- exp g 10 + ln PMSi +1( )g 12( )eacute

eumlugraveucirc( )acute100

u1i

28

Exhibit 6 Number of homicides in El Salvador by month with ARIMA Model Fit

Exhibit 5 Change in homicides from pre-truce to post-truce

Obs Mean SD 95 confidence interval

Pre-truce period 26 35442 4200 33745 37138

Post-truce period 28 21835 6361 19369 24302

Change -13606 10638 16574

=p lt 05

29

Exhibit 7 Forecast of homicides without gang truce

Next as seen in Exhibit 8 we examined the spatial distribution of the change in the homicide

rate after the implementation of the gang truce The analysis showed that of the 252 analyzed

municipalities 243 (93) experienced a decrease in homicides however within these

municipalities there were wide variations in the degree of the decline For example of the 243

municipalities that experienced a decrease the decrease in the homicide rate varied from

about 59 percent of these municipalities experiencing a 1-74 percent decrease to about 9

percent of municipalities experiencing a 75 percent or higher decrease Additionally a modest

number (n=19 7) of municipalities experienced an increase in their homicide rate

30

Exhibit 8 Percent reduction in homicide rate by number of municipalities

Exhibit 9 presents six negative binomial models for the monthly number of homicides The first

model examines the impact of the gang truce and implementation of the Violence Free

Municipality program and our interaction variable that measures the additive effect of both

the gang truce and the Violence Free Municipalities program on the number of homicides We

found that although the gang truce was associated with a significant decline in homicides the

Violence Free Municipality program was related to a significant increase in homicides13

However countrywide the additive effect of implementing the gang truce and the Violence

Free Municipalities program was associated with a significant decline in homicides

13 To be clear our bivariate analyses showed that the violence free municipalities program was related to a significant decline in homicides For example on the one hand those municipalities that did not participate in the violence free municipalities program experienced a decline in their homicide rate from 406 homicides per 100000 population in the pre-truce period to 321 homicides per 100000 in the post-truce period On the other hand those municipalities that participated in the violence free municipalities program experienced a substantially greater decline in their homicide rate from 752 homicides per 100000 population in the pre-truce period to 392 homicides per 100000 in the post-truce period However after other variables are controlled for in our negative binomial models we found that the violence free municipalities program did not have a positive impact on homicides over and above the gang truce itself

31

Exhibit 9 Negative Binomial models for monthly number of homicides

Model 1 Coefficient (se)

Model 2 Coefficient (se)

Model 3 Coefficient (se)

Model 4 Coefficient (se)

Model 5 Coefficient (se)

Model 6 Coefficient (se)

Homicide - 1 month lag 0010 0010 0010 0010 0010 0010

(0002) (0002) (0002) (0002) (0002) (0002)

Homicide - 2 month lag 0012 0013 0013 0013 0013 0013

(0002) (0002) (0002) (0002) (0002) (0002)

Homicide rate spatial lag 0003 0003 0003 0003 0003 0003

(0001) (0001) (0001) (0001) (0001) (0001)

Gang truce implemented -0483 -0541 -0553 0057 -0544 0564

(0066) (0065) (0064) (0278) (0064) (0279)

Violence free municipality 0497 0183 -0003 -0076 -0037 -0094

(0181) (0167) (0152) (0143) (0153) (0145)

Ln( +1) of MS13 prisoners 0066 0176 0062 0171

(0023) (0028) (0043) (0028)

Ln( +1) of 18th St prisoners 0067 0044 0064 -0043

(0021) (0024) (0021) (0024)

MS13 prisoner spatial lag -0004 0054 -0003 0049

(0045) (0055) (0046) (0056)

18th St prisoner spatial lag 0022 0014 0019 0012

(0027) (0032) (0027) (0032)

Gang truce violence free municipality

-0346 -0294 -0309 -0161 -0304 -0059

(0158) (0156) (0158) (0146) (0158) (0145)

Gang truce Ln( +1) of MS13 prisoners

-0185 -0185

32

(0030) (0030)

Gang truce Ln( +1) of 18th St prisoners

0160 0060

(0025) (0025)

Gang truce Ln( +1) of MS13 prisoners spatial lag

-0139 -0138

(0064) (0064)

Gang truce Ln( +1) of 18th St prisoners spatial lag

0023 0023

(0039) (0039)

Ln( +1) of MS13 on street 0023 0018 0021 0016

(0013) (0012) (0013) (0012)

Ln( +1) of 18th St on street 0043 0039 0040 0037

(0016) (0015) (0016) (0015)

Time 0010 0011 0011 0011 0011 0011

(0002) (0002) (0002) (0002) (0002) (0002)

1month 0064 0065 0067 0067 0067 0067

(0055) (0055) (0055) (0055) (0055) (0055)

2month -0020 -0020 -0022 -0023 -0022 -0023

(0056) (0056) (0056) (0056) (0056) (0056)

3month 0088 0088 0188 0088 0088 0088

(0052) (0052) (0052) (0052) (0052) (0052)

4month -0011 -0011 -0011 -0011 -0011 -0011

(0052) (0052) (0052) (0052) (0052) (0052)

5month 0020 0020 0020 0021 0020 0021

(0052) (0052) (0052) (0052) (0052) (0052)

6month 0000 0000 0000 0000 0000 0000

7month 0036 0036 0036 0036 0036 0036

(0052) (0052) (0052) (0052) (0052) (0052)

33

8month 0047 0047 0047 0046 0047 0046

(0051) (0051) (0051) (0051) (0051) (0051)

9month -0055 -0055 -0057 -0058 -0057 -0058

(0053) (0053) (0053) (0053) (0053) (0053)

10month 0040 0040 0041 0041 0041 0041

(0052) (0052) (0052) (0052) (0052) (0052)

11month 0022 0022 0022 0022 0022 0022

(0052) (0052) (0052) (0052) (0052) (0052)

12month -0040 -0040 -0042 -0043 -0042 -0043

(0053) (0053) (0053) (0053) (0053) (0053)

Urban opportunity factor 0209 0042 0033

(0040) (0040) (0039)

male 10-29 0039 0007 0009

(0021) (0019) (0019)

female-headed household -0006 0002 0000

(0007) (0007) (0006)

unemployed 0001 -0002 -0000

(0006) (0005) (0005)

Racialethnic heterogeniety 0058 -0039 -0087

(0263) (0235) (0230)

Intercept -10571

-10544

-11054 -11618 -11015 -11558

(0064) (0060) (0194) (0241) (0212) (0257)

Ln(Alpha) -2287 -2289 -2293 -2294 -2293 -2294

(0113) (0113) (0114) (0114) (0114) (0114)

Truce random effect 0202 0193 0202 0144 0201 0143

(0039) (0038) (0040) (0031) (0040) (0031)

Intercept random effect 0315 0226 0192 0164 0183 0158

(0047) (0035) (0032) (0027) (0031) (0027)

Cov(Truce and Intercept) -0162 -0099 -0122 -0086 -0114 -0090

(0037) (0031) (0031) (0025) (0031) (0026)

34

The second model includes those variables from Model 1 but also includes variables that

controlled for community-level structural factors In this model the gang truce remains

significantly associated with a decline in homicides but the Violence Free Municipality program

and our interaction variable were no longer related to a reduction in homicide Although the

percentage of residents in a municipality who are male aged 10 to 29 female-headed

households percent unemployed and ethnic heterogeneity were unrelated to changes in

homicide the urban opportunity factor was significantly associated with homicide

Models 3 and 4 included our measures of intervention as well as our measures of the number

of MS13 and 18th Street on-the-street and incarcerated gang members at the municipality level

The analysis showed that while the number of MS13 on the street was unrelated to homicides

the number of 18th Street gang members was associated with an increase in homicide These

analyses also indicated that the number of incarcerated MS13 and 18th Street gang members

from a municipality was associated with a significant increase in homicides We further

examined whether this relationship was associated with the gang truce by including two

interaction variables as presented in Model 4 One measured the interaction between the

number of incarcerated MS13 gang members and the gang truce and a second measured the

interaction between the number of incarcerated 18th Street gang members and the gang truce

The results showed that following the gang truce the number of incarcerated MS13 gang

members was related to a significant decrease in homicides and the number of incarcerated

18th Street gang members was related to a significant increase in homicides

Models 5 and 6 in Exhibit 9 present the same two models (3 and 4) as above but include the

community-level structural variables Model 5 once again shows that the number of MS13

members on the street remains unrelated to the change in homicides and the 18th Street

members on the street was associated with a significant increase in homicide However Model

6 shows that following the gang truce the number of incarcerated MS13 gang members

remained significantly associated with a decline in homicides while the number of incarcerated

18th Street gang members remained significantly associated with an increase in homicides

35

Community-level structural factors were found once again to be unrelated to changes in

homicide rates

Exhibit 10 below further illustrates the relationship between reductions in homicides by

municipality and the municipal-level presence of MS13 and 18th Street gang members in

prison The figure shows that in municipalities with no incarcerated MS13 and 18th Street gang

members no change in homicides occurred following the gang truce However when a

municipality had ten MS13 gang members imprisoned on average that municipality

experienced a 55 percent reduction in homicides When a municipality had fifty MS13 gang

members imprisoned those municipalities on average experienced a 71 percent reduction in

homicide following the truce Conversely our analysis showed that the number of 18th Street

gang members in prison (from a municipality) had a significant and positive impact (that is the

number increased) on homicides following the truce For example if a municipality had ten

18th Street gang members imprisoned on average that municipality experienced a 31 percent

increase in homicides14

14 We examined whether there was an interaction effect between gang members on the street following the gangs truce Analysis for both MS13 and 18th street and their interaction with the gang truce showed no effect

36

Exhibit 10 Impact of the gang truce by number of gang members in prison at the municipal level

The Gang Truce and Crime Displacement

We examined two types of displacement crime type displacement and method displacement

Some policymakers have suggested that although homicides may have declined as a

consequence of the gang truce other forms of criminality or categories of reported crime may

have subsequently increased Crime type displacement occurs when offenders who focus on

one type of crime switch to another form of crime in order to avoid detection or to benefit in

some other way (Eck 1993) In El Salvadorrsquos case some critics of the gang truce have suggested

that as homicides decreased other forms of criminality such as extortion might have

increased substantially due to gang membersrsquo increased freedom to conduct activities inside

and outside of prison (Dudley 2013 Parkinson 2014) Method displacement occurs when

offenders change their tactics or methods of conducting crimes as a consequence of an

intervention (Eck 1993) Following the gang truce in El Salvador some analysts argued that gang

37

members might have begun to hide the bodies of homicide victims to avoid detection and to

ldquoprotect the integrity of the countryrsquos gang trucerdquo (Bargent 2013 1)

As noted above we rely on police data that measured monthly numbers of thefts extortions

robberies rapes and auto theftsrobberies by municipality We also received data on the

annual number of disappearances in each municipality by year Exhibit 11 presents the

descriptive statistics for these variables It shows that the rates of theft and robbery did not

change significantly between the pre- and post-true periods and the rate of extortions

significantly declined The t-tests showed that the rates of rape disappearance and

theftrobbery fromof an auto increased significantly

Exhibit 11 Descriptive statistics for measures of displacement

Mean Std Err [95 Conf Interval]

Theft rate

Pre-truce 1238 049 1142 1335

Post-truce 1213 049 1115 1310

Extortions rate

Pre-truce 353 023 308 398

Post-truce 271 021 229 314

Robbery rate

Pre-truce 544 025 495 593

Post-truce 553 025 504 602

Rape rate

Pre-truce 049 003 043 056

Post-truce 064 005 054 073

TheftRobbery Vehicle rate

Pre-truce 132 022 090 175

Post-truce 165 023 120 210

Disappearance rate

Pre-truce 805 068 670 940

Post-truce 1007 064 882 1132

p lt 05

We further examined the data (excepting disappearance data) similarly to the analyses above

in which we used random effects negative binomial models and regressed the number of

38

crimes (ie theft extortion robbery rape and auto theftrobbery) on various months on

predictors with both temporal and spatial lags Our independent and control variables

remained the same as those used in Model 6 (displayed above in Exhibit 9) We found that

over the study period there was no significant change between the pre-truce and post-truce

periods in the number of thefts extortions robberies rapes and auto theftsrobberies (tables

not shown)

As noted above we also received data on the number of disappearances by municipality and

year but because the data were provided by year there were not enough data points to

examine them temporally Therefore we added this covariate (number of disappearances by

municipality and year) to Model 6 in Exhibit 9 The results of the analysis indicated that the

number of disappearances was not significantly related to change in homicides the gang truce

remained associated with a significant reduction in homicides and our interaction variables

(number of incarcerated MS13 amp 18th Street members truce) remained significantly

associated with homicides

Conclusions

The present study sought to examine the impact of the gang truce on violence and other forms

of criminality We examined homicide data by municipality prior to and after the gang truce

Our outcome variables were obtained from the PNC along with several independent and

control variables obtained from the Ministry of Economy and the National Bureau of Prisons

We analyzed this data using a number of time series and random effects negative binomial

models where we regressed the rate of homicide on various months on predictors with both

temporal and spatial lags and controlled for other contextual factors This type of analysis

allowed us to correct for the number of homicides in one month being related to the number of

homicides in the previous month (ie temporal autocorrelation) Additionally it allowed for

the fact that some communities participated in supplemental interventions (ie violence free

municipalities initiative) which might have impacted the outcome in the same way in a given

month These techniques allowed us to isolate the effects of the gang truce as best possible

We also used data from the PNC that measured by municipality the number of thefts

39

extortions robberies rapes auto theftsrobberies and disappearances to examine the

possibility of the trucersquos impact on crime displacement and diffusion of benefits`

Our findings suggest that El Salvadorrsquos gang truce had a significant and dramatic impact on the

homicide rate The mean number of monthly homicides declined from about 354 prior to the

truce to about 218 following the truce for a net decrease of about 136 homicides per month

Our forecast showed that between March 2012 and June 2014 the truce had saved about 5501

lives As will be discussed further below the analysis suggests that key stakeholders have the

capacity to renegotiate existing norms of violence and that at least some gangs have the

capacity to exert substantial informal social control over their members that can result in

reduced violence

The gang truce also lasted substantially longer than previously evaluated truces Ordog et al

(1995) for example reported that the much publicized gang truce in Los Angeles decreased

homicides by about 35 percent for the first three months but then doubled in months four

through eleven compared with the pre-truce period Similar findings were reported in Trinidad

and Tobago (Maguire Katz and Wilson 2013) where it was determined that homicides

declined for a brief period of time (again for about three months) but then increased

substantially over the long term (12 months) These results suggested that gang truces may

produce short-term benefits yet result in long-term adverse consequences

The findings of the present case study suggests that some gang truces might last longer than

previously believed While the number of homicides began to slowly increase about 12 months

following the truce in El Salvador the results clearly showed that almost two years following

the truce homicides still remained below those experienced prior to the truce

We found however that the gang truce did not result in a homogenous decline in violence

across municipalities About 61 percent of municipalities experienced a decline in homicides

but the decline in violence varied substantially between municipalities For example while

about 16 percent of these municipalities experienced a 25 percent reduction in homicides a

number of others (37) experienced a 75 percent or greater reduction Furthermore it is

important to note that about 20 percent of municipalities experienced a modest increase in

40

homicides This suggests that the impact of a gang truce might be variable and could be

dependent on contextual factors We examined the possible influence of these factors by

assessing the impact of social structural factors and the presence of gangs on the municipal-

level impact of the gang truce While we found that social structural factors were unrelated to a

decline in homicides municipal-level gang presence was associated with the decline in violence

as a consequence of the gang truce

We examined this issue further by parsing out the relative influence of the number of MS13

and 18th Street gang members on the street and in prison from each municipality As noted

above we hypothesized that municipalities where gang member presence was high regardless

of their presence on the street or in prison would experience a greater reduction in homicides

because of their increased influence in these areas We found however that the relationship

was not as direct as we would have suspected In particular our analyses indicated that

following the truce the number of MS13 and 18th Street gang members on the street in a

municipality was not significantly related to a decline in homicide but the number of

imprisoned MS13 and 18th Street gang members was associated with a significant change in

homicides following the gang truce

Of special note was our finding of the differential impact of the truce based on gang Although

we found that the number of imprisoned MS13 gang members was associated with a significant

decline in homicides following the gang truce we also found that the number of imprisoned

18th Street members was associated with a significant increase in homicides following the truce

These divergent findings might be associated with each gangrsquos organizational structure and its

capacity to regulate member behavior

Much prior research suggests that imprisoned MS13 gang members have substantial influence

over violence in Salvadoran communities15perhaps even more so than formal mechanisms of

15 Instituto Universitario de Opinioacuten Puacuteblica (Iudop) La situacioacuten de la seguridad y la justicia 2009-2014 entre expectativas de cambio mano dura militar y treguas pandilleras Aguilar Jeannette (Coord) [et al] --1ordf ed --San Salvador El Salv Instituto Universitario de Opinioacuten Puacuteblica (Iudop) 2014- 260 p graacuteficos tablas cuadros y figuras 28 cm (Talleres graacuteficos UCA) httpwwwucaedusviudopwp-contentuploadslibro_la_situaciC3B3n_de_la_seguridadpdf

41

social control such as the police and courts16 Law enforcement officials for years have claimed

that MS13 is one of the most organizationally sophisticated street gangs in the Western

Hemisphere17 The gang has been widely characterized as having a highly vertical organizational

structure and strong control over criminal enterprises in gang-controlled neighborhoods and as

being decidedly capable of enforcing rules through discipline MS13 leadership resides in the

nationrsquos prison system The Ranfla (gang leadership) is comprised of thirteen MS13 gang

members (PNC 2011) who direct coordinate and authorize street crime and other gang activity

from prison Our findings lend support to the idea that MS13 is more organized than the typical

street gang and that imprisoned MS13 gang members exhibit strong influence over their fellow

gang members on the street

Our findings also suggest however that the gang truce had a boomerang effect in

municipalities with high numbers of imprisoned 18th Street members implying that 18th Street

might not have as much organizational capacity to regulate violence on the streets as MS13

The truce provided incarcerated MS13 and 18th Street gang leaders an opportunity to

negotiate with high-ranking officials and influential diplomats This may have increased their

legitimacy inside and outside of their gangs It appears that MS13 was able to exert its span of

control over the communities in which they had influence and they were able to deliver on the

terms of the gang truce negotiations In the case of 18th Street however incarcerated gang

members may not have had the same organizational capacity for communicating and carrying

out directives In fact a review of the gang truce indicated that there was a conflict taking place

between two fractions within 18th Street Consequently the organizational structure and

culture of 18th Street might be more diffuse than that of MS13 and its leadership structure

might not be as strong because of the internal fractures within the gang This might further

explain why homicides increased in 18th Street communities The internal fractures within the

18th Street gang may have resulted in intra-gang violent conflict that was largely contained

16 Instituto Universitario de Opinioacuten Puacuteblica (Iudop) La situacioacuten de la seguridad y la justicia 2009-2014 entre expectativas de cambio mano dura militar y treguas pandilleras Aguilar Jeannette (Coord) [et al] --1ordf ed --San Salvador El Salv Instituto Universitario de Opinioacuten Puacuteblica (Iudop) 2014- 260 p graacuteficos tablas cuadros y figuras 28 cm (Talleres graacuteficos UCA) httpwwwucaedusviudopwp-contentuploadslibro_la_situaciC3B3n_de_la_seguridadpdf 17 httpwwwlaprensagraficacomeua-declara-a-ms-grupo-delictivo-transnacional

42

within 18th Street controlled territories Further analysis is needed to examine this specific

issue

While not the primary focus of this case study we controlled for any impact that the Violence

Free Municipalities program might have had on homicide in 11 municipalities during the truce

As noted above the Violence Free Municipalities program served as the second phase of the

gang truce (Cawley 2013) The program was first proposed by the two mediators and designed

by the Technical Committee on Violence and Crime Reduction which included representatives

from OAS MJPS the mediators and the Humanitarian Foundation (CISPES 2013) The program

involved the mayors of each municipality collaborating with gang leaders to design prevention

and intervention resources for gang members and at-risk youth The Minister of Justice and

Security David Mungia Payes announced that his ministry would facilitate $74 million in

funding from the OAS UN and other donors to implement the programing In exchange gang

leaders agreed to end violence and other criminality in the Violence Free Municipalities (CISPES

2013) The negotiators also agreed to further discuss both gangsrsquo demand to repeal of the

2010 law that increased the capacity of the police and prosecutors to crack down on gangs

(Ayala 2014)

Our findings suggested however that the Violence Free Municipalities program was unrelated

to change in homicides in these communities While much additional research is needed to

understand why the program was not effective it might have been because the gangs had

already agreed to a truce and had already reduced violence to the extent that they could

Conversely the demands made by the gangs as part of the Violence Free Municipalities

program might have been more than could be delivered The time and resources required to

implement the programming and the political capital that was required to repeal legislation

might have been much more than could be delivered Future research is needed to examine the

processes and impact of the Violence Free Municipalities program

Regardless the base rate of violence in El Salvador was reset for a period of time suggesting

that perhaps the gang truce substantially altered existing norms of violence Klein (1995)

argued that cycles of gang violence (that is perceived or real changes in gang activity) are

43

typically the consequence of seasonality epochal variation (peaks and valleys in violence) and

illusion (it appears as if there is a real change in violence but there is not) In El Salvador we

appear to have observed a self-directed cycle of normative change wherein incarcerated MS13

gang leaders directed a reduction in violence by actuating their organizational span of control

through the gang truce Some of the most powerful and influential gang members in El Salvador

used their political social and economic capital to promote the truce and articulate new norms

of violence They were able to do this by leveraging their informal social control over the streets

through actual or perceived threats of violence against those who violated the terms of the

truce

However following a change in government leadership and the government subsequently

distancing itself from the gang truce the conditions and capacity of MS13 leaders to intervene

in local violence might have deteriorated and violence began to increase substantially This is in

part because the third parties in the negotiation were no longer able to communicate with

government officials about furtherance of the truce Specifically the mediators were no longer

able to negotiate on the behalf of the government and were no longer permitted entrance to

the prisons where they could negotiate with gang leaders

44

Case Study B Gang Trucemdash The Jamaican Experience

In the latest report by the UN Office of Drugs and Crime Jamaica ranked as the sixth most

violent nation in the world with a homicide rate of about 521 per 100000 residents (UNDOC

2011) Policymakers and researchers have attributed the nationrsquos high level of violence to such

factors as drug trafficking (Klein Day and Harriott 2004) access to illegal firearms (Lemard and

Hemenway 2006) and historical processes that include a legacy of conflict between the

nationrsquos two primary political parties (Sives 2002 Figueroa and Sives 2003 Moser and Shrader

1999) all of which have facilitated the entrenchment of the more powerful gangs in

communities of the urban poor Likewise high levels of income inequality and chronic youth

unemployment (Francis and Lyare 2006) problematic urbanization (Stone 1975) social

marginalization and an emergent subculture of violence (Harriott 2008) have contributed to the

nationrsquos violence problem Whatever the causes violence has had a considerable impact on

Jamaicarsquos social and economic development it has decreased investor confidence (Schwab and

Porter 2008) tourism (Harriott 2007) and access to public services It has also increased the

costs related to the health care system (Mansinghand and Ramphal 1993) the criminal justice

system (Caribbean Human Development Report 2012) and the education system (Moser and

Holland 1997)

Jamaicarsquos homicide problem is closely associated with its gang problem The Jamaican

Constabulary Force (JCF) has estimated that some 272 gangs are active in the nation most in or

near the capital city of Kingston (Harriott 2014) Gang types and their respective historical

patterns of conflict matter in Jamaica as these variations may determine their predisposition or

amenability to lasting rather than opportunistic truces Jamaican gangs include territorially

organized crime groups conflict gangs defense crews who regard themselves as defenders of

their communities (Levy 2009) and other less cohesive more transient territorial groups Some

of the latter are predatory others bond around the taken identities that generate conflicts of

other similar groups18

18 These are generally rooted in subcultural issues such as the demand to be treated with respect or with the indiscretions and self-centered aggressiveness of individual members that may be associated for example with sexual competition for the favors of women

45

Estimates of the gang-related homicide rate in Jamaica vary perhaps because the crime is not

clearly defined as far as attributing a death to a gang Regardless researchers agree that the

proportion of homicides that are gang related has increased substantially For example Harriott

(2003) reported a fourfold increase in the rate of gang-related homicides between 1983 and

1997 Likewise Hill (2013) using official police data found an eightfold increase from 2001 to

2009 reportedly in 2001 only about 64 percent (n=73) of the nationrsquos 887 homicides were

gang related compared with 523 percent (n=879) of its 1682 homicides in 2009

To address the problem Jamaica has initiated traditional law enforcement strategies such as

establishing a specialized gang unit (Sinclair 2004) initiating curfews (Sinclair and Tuner 2005)

declaring states of emergency (Jamaican Observer 2010) and implementing community-

oriented policing (Kolpack 2006) It also attempted legislative reforms to curb election fraud

and electoral-related violence that involved local gangs (Levy 2009) None of these strategies

stemmed the tide of gang violence In 2002 the Minister of National Security established the

Peace Management Initiative (PMI) (Henry 2011) to augment governmental and non-

governmental organizational capacity to settle gang disputes in the community through

intervention-based programming such as ceasefires and gang truces

The current study examines the peace initiative instituted by the PMI in Greater August Town

Our objective was to understand the negotiation processes undertaken with and between

gangs and other stakeholders Among other things we were interested in identifying the actors

involved in the negotiations the goal(s) of the negotiations and the strategies employed to

carry them out Most importantly however we wanted to determine whether the gang truce

resulted in the desired outcome a reduction in the number of homicides in the Greater August

Town area

The Greater August Town (Jamaica) Peace Initiative

Greater August Town is located on the northeastern outskirts of the city of Kingston The low-

income area has high rates of youth unemployment and a history of gang-related violence

(Charles 2004 Levy 2009) Over the last decade the Greater August Town community has

sought improved living conditions and revitalization (Levy 2009 95) The arearsquos inherent

46

resilience has been augmented by nearby intellectual and cultural engines such as the

University of West Indies the University of Technology and University Hospital (Charles 2004

38)

Greater August Town is comprised of the communities of August Town proper (which is

fractured into several locales with gang-given names such as ldquoVietnamrdquo and ldquoOpen Landrdquo and

city government-approved names such as Hermitage Goldsmith Villa (Angola) and Bedward

Gardens These socially defined community divisions and subdivisions are markers for the

territorial boundaries of street gangs and therefore in some instances are lines of potential

conflict Some of those boundaries demarcate areas of Greater August Town that are

predominantly supportive of one or another political party but the boundaries do not always

hold political significance politics is but one element in the conflict geography of the area Like

many communities of the urban poor the Greater August Town area is easily mobilized

politically ndash a reality that is understood and at times exploited by street gangs who politicize

gang ldquowarsrdquo in their efforts to build alliances and to neutralize the police In fact the basic

principle of community mobilization in Jamaica is political patronage and clientelism Access to

resources (eg jobs housing education) for the poorest residents is often determined by the

local political party Thus according to Charles (2004 36)

supporters attach themselves to the political parties to get first preference in the

distribution of scarce resources and over time because they are unemployable they

become dependent on their political party for their economic survival These supporters

will kill anyone who threatens the support base of their political party because they

perceive it as a threat to their daily survival

As a consequence some political supporters invest heavily in the electoral contests and

provoke conflicts that affirm their loyalty to their party in order to secure material benefits

from it Political competition is one conflict fault line in what otherwise is a politically

heterogeneous community Specifically political support in Greater August Town is divided

between the Peoplesrsquo National Party (PNP) and the Jamaica Labor Party (JLP) The PNP receives

strong support from residents in August Town proper the upper region of Goldsmith Villa and

47

Bedward Gardens while the JLP is supported by those living in Hermitage and the lower region

of Goldsmith Villa (Charles 2004 ) The division between the political parties in the area as in

much of Jamaica was believed to be at the root of much of the communityrsquos violence

particularly between 1979 and 1993 Some gangs aligned with the PNP and others with the JLP

this often resulted in political boundaries overlapping with gang turf (Charles 2004)

Pre-truce Violence

Nationally the history of gang violence in the Greater August Town area first appeared as

political violence closely associated with the electoral cycle As in many other urban

communities the problem intensified mdash deeply affecting community life in the period just prior

to the national elections of 1980 mdash then continued cyclically until 1993 Gangs have since

harnessed this legacy in order to establish and maintain community support based on common

political affiliations They have used that support to nullify the efforts of law enforcement to

suppress their illicit activities (Harriott 2008)

The conflict profile of these gangs and of the communities in which they are nested has

changed over the decades From the beginning of their involvement in political violence and

territorial control a form of gerrymandering existed that manipulated the voting population

forcing certain individuals out of a particular constituency and preventing those who remained

in the community from voting for the opposing party This was linked both to the electoral cycle

and to the types of inter-gang conflicts that typically were associated with street gangs

(Figuerou Harriott and Satchell) More recently however much of the crime and violence

perpetrated by gangs has resulted from internal conflicts (eg status management disputes

over womengirls or money individual membersrsquo activities that could attract police pressure)

Internal conflict at times has led to gang fragmentation and new alliances that pull more parties

into the conflict escalating the homicide rate and increasing the sense of insecurity among the

general population (Levy 2012)

The most significant of these internal conflicts resulted from the killing of former Jungle 12

leader Neil Wright by members of his own gang Jungle 12 was the most influential gang in

Greater August Town Before his murder in order to increase the gangrsquos access to illicit

48

opportunities in Kingston Wright had been attempting to extend Jungle 12rsquos influence with a

system of alliances with other gangs and recruitment in Kingston (Harriott 2014) In short his

ambition was to transform Jungle 12 from a neighborhood street gang operating on the

outskirts of Kingston into to a dominant organized crime network that could reach into the

heart of the city In pursuit of this goal Wright recruited members from outside August Town

elevating them in the gang hierarchy above the locals This led to status-related conflicts and

resistance to Wrightrsquos leadership within the gang His murder precipitated a split of Jungle 12

into three factions two of them fled to other neighborhoods within August Town (Angola and

Vietnam) resulting in the formation of new alliances and a new conflict geography that

replaced the former political geography of conflict Wrightrsquos killing and the subsequent demise

of Jungle 12 as the dominant gang in Greater August Town altered the balance of power and

escalated inter-gang violence (Harriott 2014) The post-2005 phase of conflict was

characterized by power symmetry conflict intensification and the spread of conflict

throughout the entire geographic area of Greater August Town

Although their origins are unclear retaliatory killings and other violent incidents progressively

intensified between 2005 and 2008 The violence was episodic retaliations were most often

motivated by suspicions related to personal and geographic connections between warring

gangs As the violence escalated new alliances were formed to enhance power and dominance

which in turn increased the number of gangs and gang members involved in the violence

(Harriott 2014) This eventually attracted national attention and triggered community

mobilization for a gang truce

The Truce-making Process

The Greater August Town gang truce was preceded by frequent intense violence and public

outrage As noted above the violence had escalated in November 2005 when Jungle 12 leader

Neil Wright was killed The defection of a Jungle 12 member to Goldsmith Villa (Angola) caused

infighting within the gang and conflict between it and Goldsmith Villa Just a few months later

Wrightrsquos brother Steve and two others were injured during a turf battle (Martin-Wilkens 2006)

49

Thereafter violence began to occur at regular intervals until January 2007 when the Peace

Management Initiative (PMI) hosted a peace march in the community Two PNP politicians

urged the community to unite A PMI leader declared that the peace march was being held to

ldquodemonstrate to the public that Jungle 12 members are back together and that they want

peacerdquo (Thompson 2007 1)

Although hopeful some residents remained skeptical about the peace march perceiving the

action to be politically motivated In the absence of trustworthy information inter-group

conflicts tended to be interpreted through a politically partisan lens this created obstacles to

isolating the gangs building a consensus for peace and unifying community mobilization The

politically based narratives weakened the communityrsquos leverage for peace as well as the

exposure of the gangs to police action As one resident said ldquoThe election is coming up and

they want[ed] the people to vote for the PNP is one of the main reasons why they have to walk

todayrdquo (Thompson 2007 1) Those who shared such views stayed away from the peace march

Although that widely held myth was not factual it did serve to demoralize and demobilize one

part of the community A local UWI faculty member was articulate in his assessment of the

politics behind the march

August Town violence is not violence of organized crime which is based on drug

trafficking extortion or some other criminal enterprise [but] rather the violence in

August Town is essentially ldquotribalrdquo mdash the Peoplersquos National Party tribe versus the

Jamaica Labour Party tribe [which has been complicated by a splintering within the

PNP tribe] (2007 1)

Indeed it was not the violence of organized crime but neither was it political violence The

individual quoted above neglected to mention that the conflict was between those pro-PNP

splinter groups who were largely comprised of members of Jungle 12 Moreover their ldquopro-

PNP-nessrdquo was unrelated to the conflict there was no factional infighting within the local PNP

organization at that time

Nevertheless following the peace march the gang violence diminished Then in November

2007 a turf war erupted between two gangs from the Greater August Town neighborhoods of

50

Vietnam and River This time as the police stated the gang violence was less about politics and

more about dominance and turf Police were dispatched to perform directed patrols but

whenever they were not present the shootings continued (Mcleod 2007) In April 2008 the

community witnessed local gangs engaging in a five-hour-long street battle that left two killed

and three others wounded It ended only after the police deployed armored trucks The next

month another round of gang violence resulted in five others being killed including a one-year-

old child This resulted in the three members associated with the gang who committed the

homicides being killed in retaliation (Virtue 2008) The local community mobilized against the

violence increasingly cooperating with the police providing more information about the gangs

Subsequently gang members observed a decline in their influence within the community

During the early period characterized by low-intensity conflict the less influential gangs at

times used manipulation of the police as a tactic for suppressing the more influential gangs

This was largely done through strategic release of information Prior to 2005 when Jungle 12

was dominant its membersrsquo illicit activities were constantly reported to police by members of

other gangs as a means of compelling a compromise or settlement of conflicts In practice this

was done by ldquotrading casesrdquo Once a crime had been investigated by the police and suspects

had been charged an opportunity was created for the gangs and other parties to the conflict to

settle the matter by agreeing to drop their cases (typically by ceasing cooperation with police

investigators) This type of ldquoself-helprdquo served to end some of the retaliations but it rested upon

the manipulation of the police (Harriott 2014)

Later in an attempt to quell escalating inter-gang violence the police established buffer zones

between the warring gangs This action resulted in unintended consequences For example

when the police declared a buffer zone between August Town and Hermitage Hermitage took

advantage of the opportunity to attack Angola Some Angola residents accused the police of

turning a blind eye and creating an opportunity for Hermitage to attack their community

Although little reliable information exists about why the police made the deployment the way

that they did it is more likely that the police inadequately assessed the situation (ie mis-

assessed the pattern of alliances and the likely targets of attack) (Harriott 2014) In the areas

affected by this kind of increasing violence community members became angered and lost

51

confidence in the police The error resulted in some parties to the conflict receiving increased

support from their communities and in greater gang-community cohesion (Harriott 2014)

After a brief period the police identified this problem and began to disengage by no longer

providing a buffer between gang controlled areas which in turn allowed still more conflict to

occur between the gangs

As the violence escalated beyond their control police finally responded by applying their own

forms of pressure For example units under the direction of the JCF High Command would

make periodic raids in the community during which they would at times seize weapons and

make mass arrests (Sinclair 2005) However there were also moments when the local police

were very responsive improved their relations with the community and consequently gained

greater access to relevant information Two such moments occurred just prior to and again

immediately after the truce moments during which there was greater freedom of movement

and open collaboration between the community and the police (Harriott 2014)19

The Establishment of the Greater August Town Gang Truce

The Greater August Town gang truce was led by the Jamaican Peace Management Initiative

(PMI) The PMI is a government-funded initiative created for the purpose of working with gang

members to reduce violence Due to community mistrust of the police in 2002 the organization

was established as an alternative organizational mechanism for responding to gang violence

The PMI sought to bridge government and civil society efforts to mediate disputes between

gangs as well as to provide outreach to gang members (Bakrania 2013) While efforts to

institute a gang truce in Greater August Town were led by the PMI a number of other

stakeholders helped to facilitate the truce these included faculty at the University of the West

Indies (UWI) and representatives from the police the faith-based community and the August

Town Sports and Community Development Foundation (Jackson 2008 Levy 2009 also see

Appendix C) The gangs involved in the truce included those from August Town Hermitage

19 The quality of police-community relations largely depended on the style of the local station commander however regardless of the external environment

52

Goldsmith Villa Bedward Gardens and African Gardens Because of its formality as well as its

perceived effectiveness the truce signed on June 24 2008 was regarded by many as the first

of its kind in Jamaica (Levy 2009)

Truce negotiations began early in June 2008 and lasted for about three weeks The gangs

sought to leverage their violence-making capabilities and demanded payment for peace They

asked the third-party negotiators for money ldquoworkrdquo and start-up funds for proposed micro-

businesses (Wilson 2014) Those demands were rejected by the negotiators on the grounds that

the third-party institutions would not buy a peace that was intended to save the lives of those

who were making the demands Moreover if peace was to be purchased then gang conflict

could be used continuously to extract money and other benefits from negotiators The third-

party actors made some demands of their own In some quarters of the community and society

the surrender of guns was viewed as a litmus test of the sincerity of the gangs Consistent with

this the negotiators suggested that the parties to the conflict symbolically hand over one gun

each that suggestion was immediately rejected by the gang leaders These kinds of demands

from the various parties ceased after a time as they all agreed that the truce was to stand on

its own merits (Harriott 2014)

As the truce began to be committed to paper a number of stakeholders expressed concern that

their greatest risk in participating could be the potential for Jungle 12 factions to use the peace

agreement as they had in the past as a tactic to persuade their enemies to let their guard

down Others however recognized that Jungle 12 had now been weakened and that a formal

public peace agreement would be beneficial to the gang and therefore this time would be

different (Harriott 2014)

The gangs held fast to their claim that their weapons were needed for their own protection

because the police were ineffective in responding to violence in their communities (Jamaican

Gleaner 2014) It became a precondition of the truce that the gangs would not be required to

turn in their guns and other weapons (Jackson 2008) The truce agreement did specify

however that ldquoall persons are allowed to move freely across all boundaries regardless of

reputation or affiliation No gun salute or any other shooting is to take place in the community

53

for a period of at least five yearsrdquo (2008 also see Appendix C) The truce agreement and its

conditions were prescribed in a document that was finally signed by all of the major

stakeholders including the gangs (see Appendix C)

Throughout the negotiations each of the gang leaders had attempted numerous times to use

the truce as an opportunity to bargain for money jobs and business support grants Such

demands consistently were rejected by the third-party actors Nonetheless both prior to and

after the truce some efforts were made to create better opportunities for young people

residing in the community UWI for example provided a homework supervision program to

encourage students to further their education and it developed a community-building initiative

to help improve schools and enhance sports programming (Levy 2009) Such programs were

conducted as part of UWIrsquos Township Project in August Town which invested significant

resources in developing the residentsrsquo job-related capabilities and collective self-efficacy

The Greater August Town gang truce was noteworthy for two reasons First the gang truce

received substantial press attention The media were invited to witness the ldquosigningrdquo of the

truce by the gang leaders in the presence of a JCF Deputy Commissioner of Police a PMI board

member and the UWI Principal and two professors of its faculty Second the truce was widely

credited with decreasing violence in Greater August Town and it served as an exemplar to

other communities seeking to replicate its success (Virtue 2008) A number of reports

manuscripts and newspaper articles proclaimed the truce to be a success Bakrania (2013 10)

for example reported that ldquoPMI has been credited with stopping gang wars in August Town

rdquo Levy (2009 94) remarked that the ldquomost interesting outcome of PMI efforts to date was the

Peace Agreement reached in August Town in late 2008rdquo Likewise a government report noted

that ldquothe peace treaty was a pivotal achievement in August Town that has significant potential

for wider application Crime levels dropped markedly in August Town after the signing of the

peace agreement in June 2008rdquo (McLean and Blake-Lobban 2009 78) To this day August Town

celebrates the signing of the truce with an annual celebration with food and music

(Cunningham 2011)

54

Methods

Our evaluation relied on a pre-testpost-test quasi-experimental group design Our

methodology examines the Greater August Town community which is comprised of three

contiguous towns where the gang truce took place (the target area) and the balance of Jamaica

which is comprised of 178 communities (comparison areas) As seen in Exhibit 12 the average

number of residents living in each of the three communities in the target area was not

significantly different than that for the rest of Jamaica about 7776 residents lived in each of

the Greater August Town communities compared with 6468 in the other communities

Likewise communities of Greater August Town were about as densely populated as other

communities (2960 per square kilometer versus 2647 per square kilometer) and the age

range of residents was similar as well However Greater August Town (a) had a significantly

higher proportion of its residents living in poverty (196 vs 158) (b) consumed fewer

resources than other communities and (c) reported significantly more homicides than other

communities prior to the truce (see Exhibit 12)

55

Exhibit 12 Descriptive characteristics of Greater August Town and balance of Jamaica (2007-2011)

Comparison

Area Greater August

Town All areas

Population (mean) 6468 777633 648994

(sd) 720482 353731 715621

Population density (mean) 264719 296033 265238

(sd) 271023 285501 270465

Percent in poverty 1577 1957 1583

(sd) 1036 106 1029

Consumption 15737890 1106939 1566048

(sd) 10713020 205336 1064021

residents under 15 yrs old 2369 2494 2371

(sd) 487 115 484

residents 15-65 yrs old 6840 6901 6841

(sd) 423 29 419

Murder per month (mean) 674 857 677

sd 1928 1409 1920

Total murders 10068 180 10248

n 178 3 181

plt=05

Measures

Two distinct data sets were merged to measure the impact of the Greater August Town truce

First data from the 2011 decennial census provided community-level measures of the social

and economic characteristics of the 181 communities in Jamaica Described in detail below the

community-level data used in the study included population population density gender age

poverty and consumption20 These data were obtained directly from the Statistical Institute of

Jamaica

Second police homicide data from the years 2007 through 2011 were used to construct the

studyrsquos community-level measure of homicide The homicide data were aggregated by month

20 Consumption is an alternative measure of poverty in Jamaica which measures the consumption of food and non-food items

56

and appended to the community-level data The final (merged) data set included 10248

homicides over the 60-month study period These data were obtained from the Jamaica

Constabulary Force (JCF)

The dependent variable examined in the study was constructed from official police homicide

data Once again the homicide data represented the number of officially recognized homicides

in Greater August Town and each of the remaining communities in Jamaica We examined

change by comparing the homicide data 18 months prior to the truce with the homicide data 42

months following the truce More specifically we examined whether there was a change in the

number of homicides in the 30 days following the truce (month 1) as well as whether the truce

had an impact every three months thereafter (ie months 2-5 6-8 9-11 12-14 15-42) and

whether any changes in homicide coincided with changes in homicide in the balance of

observation areas The frequency distribution of our dependent variable is presented in Exhibit

13 It shows that prior to the truce the target area on average experienced significantly more

homicides (1495) than did the comparison areas (920)

57

Exhibit 13 Distribution of homicides in the target and comparison areas

Comparison

Area Target

Area Total

Pre-truce period Mean 920 1495 932

SD 2469 1966 2461

N 241400 5100 246500

Month 1 of truce Mean 741 286 733

SD 1785 496 1772

N 17800 300 18100

Months 2 thru 5 of truce Mean 647 905 652

SD 1821 1249 1812

N 71200 1200 72400

Months 6 thru 8 of truce Mean 577 1236 588

SD 1690 1074 1683

N 53400 900 54300

Months 9 through 11 of truce Mean 718 333 711

SD 2034 733 2019

N 53400 900 54300

Months 12 through 14 of the truce Mean 687 095 678

SD 1519 286 1509

N 53400 900 54300

Months 15 thru 42 of the truce Mean 564 589 564

SD 1683 1042 1674

N 516200 8700 524900

Total Mean 674 857 677

SD 1928 1409 1920

N 1006800 18000 1024800

An illustration of the trends in homicide prior to and following the gang truce are shown in

Exhibit 14 It shows that 30 days following the truce homicides fell in the target and

comparison areas then increased and decreased several times with a general downward slope

in violence over time

58

Exhibit 14 Monthly number of homicides pre-post truce in the target and comparison areas

We also used a number of measures to control for community-level structure from the 2011

decennial census These community-level data included the communityrsquos population

population density (per square kilometer) and community level of consumption Additionally

the census data included measures of the percentage of the population that was female under

15 years old 15 and 65 years old and 65 years old and older as well as a measure of the

percentage of the population living in poverty Principal components analysis was used to

reduce some of these data into a summary measure

Exhibit 15 shows the results of the component loadings One component was extracted that we

designated as socio-economic status (SES) which exhibited high loadings for percent living in

poverty percent under 15 years old percent 15 to 65 years old and consumption Excluded

from the principal components analysis were population and population density Population

was used as our exposure variable and population density was logged to address skewness in

these data

Exhibit 15 Factor loadings from principal components factor analysis

Loading Poverty 78 Consumption -76 under 15 years old 92 between 15 and 65 years old -80

59

Analytic Strategy

In order to test whether the truce had an impact on homicides in the target area andor

whether displacement had occurred in the balance of the study area several analytic

techniques were employed Most of the methods employed the use of the homicide rate as the

dependent variable We explored the data in this way to provide the maximum statistical

power to detect an effect As a check on these methods we also employed a generalized model

to compensate for the non-normality of our outcomes

First focusing only in the target area we performed a simple t-test comparing the homicide

rates before and after the truce (the unit of analysis was a month) however this technique had

limitations the most severe of which was that even if the test were significant it would be

difficult to determine whether the difference was due to the gang truce or a natural change

over time in the outcome Second to address this limitation time series models were

employed whereby the homicide rate for the target area was modeled as a function of time

with truce period indicators included to measure the effect of the truce net of the temporal

trends These models were estimated with ARIMA techniques with a one-month lag auto-

correlated error Third we examined the homicide rate for each town using a panel time series

model In this model the temporal trend for each town was examined with indicators for target

areas and truce periods included The main effects for the truce periods measured the effect of

the truce in the target areas and the moderators of the truce period in the comparison areas

measured displacement effects Finally because the dependent variable coded is not normally

distributed across months we used a negative binomial time series model to estimate the

number of homicides with the population covariate serving as an exposure variable

Findings

The first set of results examines only the target area The first test was a simple t-test

comparing the mean homicide rates before and after the truce periods The result was a mean

difference in the homicide rate of -890 per 100000 with a significant t-statistic of 370 While

60

this result is statistically significant we caution that it may or may not reflect an impact of the

truce To further examine the truce effect in the target area we performed ARIMA regressions

The first model did not include an effect of the temporal trend In this model we again find that

by month 15 the murder rate decreased by about -89 per 100000 (Exhibit 16)

Exhibit 16 Results of basic ARIMA model

Next we employed the ARIMA model again but included a variable (date) to control for the

temporal trends in the data Exhibit 17 shows that when we controlled for temporal trends the

impact of the truce we observed was no longer significant This result indicates that it was not

the truce per se that caused the decline in homicides but instead the decline in homicides was

part of a larger (local and nationwide) decline in homicides

Note The test of the variance against zero is one sided and the two-sided sigma 7877339 5907266 1334 0000 6719536 9035142 L1 1720758 1242218 139 0166 -0713944 415546 ar ARMA _cons 1538436 1904285 808 0000 1165203 1911669 t_15 -8888017 2969383 -299 0003 -147079 -3068134 t_12 -1507749 1820406 -083 0408 -507568 2060181 t_9 -132475 9189969 -144 0149 -3125951 4764507 t_6 -2861008 6335492 -045 0652 -1527835 9556328 t_2 -5166363 4498142 -115 0251 -1398256 3649834 t_1 -1111771 4463623 -025 0803 -9860311 7636769murders_rate murders_rate Coef Std Err z Pgt|z| [95 Conf Interval] OPG

Log likelihood = -2089969 Prob gt chi2 = 00397 Wald chi2(7) = 1472Sample 564 - 623 Number of obs = 60

ARIMA regression

61

Exhibit 17 Results of ARIMA model with control of temporal trends

We next estimated the possible displacement effects of the truce Exhibit 18 presents the

results of these models Examination of the main effects of the truce period does not indicate

any effects and looking at the truceComparison interaction effects we also do not find any

displacement effects Note that these models also controlled for the socio-demographic

characteristics of each community

Note The test of the variance against zero is one sided and the two-sided sigma 7775398 6622814 1174 0000 647735 9073446 L1 1270618 1253989 101 0311 -1187156 3728392 ar ARMA _cons 1480274 1197324 124 0216 -8664378 3826986 date -2318963 2091772 -111 0268 -6418761 1780835 t_15 -3519687 7694691 -005 0964 -1543329 1472935 t_12 -970747 2228007 -044 0663 -5337561 3396067 t_9 -8844212 9402792 -094 0347 -2727335 9584921 t_6 2995697 6576644 005 0964 -1259042 1318956 t_2 -2353533 4923798 -048 0633 -12004 7296933 t_1 -9194625 5462116 -017 0866 -1162501 9786088murders_rate murders_rate Coef Std Err z Pgt|z| [95 Conf Interval] OPG

Log likelihood = -2082031 Prob gt chi2 = 00710 Wald chi2(8) = 1444Sample 564 - 623 Number of obs = 60

ARIMA regression

62

Exhibit 18 Results of Panel (Town) time series model with control of temporal trends

Last we used a random effects negative binomial regression that predicted the homicide rate

with population as an exposure variable and controlled for the socio-demographic

characteristics of each community The results are presented in Exhibit 19 The analysis showed

that time had a negative effect indicating that homicides were decreasing in all areas over the

study period The main effects of the truce (truce = 1 2 ) represented the effects of the

truce in the targeted area and did not show a significant effect for any period following the

gang truce However we did find that the homicide rate significantly increased in the

_cons 6310151 3518738 179 0073 -5864483 1320675lnpopulationdensity 121961 1709831 713 0000 8844891 1554731 ses 1245334 2428629 513 0000 7693312 1721336 date -1276623 0345659 -369 0000 -1954101 -0599144 15Balance 5392222 4128863 131 0192 -2700202 1348464 12Balance 1318516 7903159 167 0095 -2304746 2867507 9Balance 1042658 7925876 132 0188 -5107854 2596101 6Balance -1703928 7919006 -022 0830 -1722489 1381704 2Balance 3498997 7175155 049 0626 -1056405 1756204 1Balance 1056874 1104696 096 0339 -1108291 3222039 truceaugust Balance -4757021 3276618 -145 0147 -1117907 1665032 august 15 -4269255 4284086 -100 0319 -1266591 41274 12 -1262496 7871907 -160 0109 -2805361 2803697 9 -1014276 7881965 -129 0198 -2559113 5305603 6 6563019 786731 008 0934 -1476334 1607595 2 -5034368 7123478 -071 0480 -1899613 8927392 1 -1215158 1095699 -111 0267 -3362688 9323715 truce murders_rate Coef Std Err z Pgt|z| [95 Conf Interval]

Prob gt chi2 = 00000 Wald chi2(16) = 12359 max = 60 avg = 5661878Estimated coefficients = 17 Obs per group min = 48Estimated autocorrelations = 1 Number of groups = 181Estimated covariances = 1 Number of obs = 10248

Correlation common AR(1) coefficient for all panels (02045)Panels homoskedasticCoefficients generalized least squares

Cross-sectional time-series FGLS regression

63

comparison areas in months 12 through 14 following the truce In particular we found a 29

percent increase in the homicide rate in the comparison communities for that period (exp (-

1797 + 2048) = 1285 p-value = 004) Since this effect is only significant at the 005 level

however and given the number of analyses used to examine the data it is possible that we

found this effect by chance alone

64

Exhibit 19 Random Effects Negative Binomial

65

Given these caveats we visualized this model with the following set of marginal predictions as

observed in Exhibit 20 We saw that the targeted area (as illustrated in red) experienced an

immediate decrease in homicide which coincided with an increase in homicides in the balance

of the study area However the target area quickly returned to ldquonormalrdquo and homicides in the

comparison area decreased again During months 9 through 11 following the truce there was a

reduction in homicides in the target area with an associated increase in the comparison area It

is important to point out that the confidence intervals are large and we cannot yield concrete

conclusions from these results However it appears that the truce might have had a temporary

short-lived displacement effect decreasing homicides in the target area but increasing

homicides in the comparison area

Exhibit 20 Predicted change in homicides in the target and comparison areas

Conclusions

From 2000 through 2009 Jamaica experienced a substantial number of homicides many of

which were attributed to gangs in one form or another Traditional law enforcement responses

were repeatedly implemented but until 2010 those had little effect Some policymakers in

Jamaica as well as in other nations throughout the Caribbean and Central America have

recently been experimenting with novel approaches to reducing gang-related violence notably

the implementation of gang truces In Jamaica at least eight gang truces reportedly have been

66

negotiated since 2001 (Levy 2009) The Greater August Town gang truce was thought to have

been one of the more successful and it has served as a model for other communities to use

(2009) Our purpose here has been to identify the actors involved in the negotiations of that

truce the negotiation goals and the implementation methods used and then to examine

empirically the impact of that truce on homicide rates in the targeted community

The 2008 gang truce in August Town was a response to violence that arose when the leader of

one gang was killed creating a power vacuum that other gangs saw as an opportunity to

increase their influence in the community Concomitantly the community as well as the gangs

feared that an absence of formal social control would result in further violence The police

reacted unevenly At some times they engaged in appropriate but heightened levels of

preventive patrol while at other times they purposely provided little or no protection on

occasion they used aggressive tactics that further isolated them from the community The end

result was that there was neither stability nor predictability in the police response and

therefore little trust in the police to address the problem

As the violence further escalated the community mobilized The Jamaican Peace Management

Initiative faculty members from the University of the West Indies (UWI) the Jamaican

Constabulary Force (JFC) and other community-oriented groups joined forces seeking to

reduce the increasing number of homicides by brokering a truce between the gangs Over the

three-week negotiation period the negotiators and the gangs sought terms from one another

The gangs wanted payment ldquoworkrdquo and funds for micro-business development to end the

violence The third-party stakeholders wanted the gangs to disarm actually or symbolically

Neither the gangs nor the stakeholders had substantial leverage nor did they have much to

offer one another in terms of incentives In the end however a truce was agreed upon and all

of the gangsrsquo leaders and several key community stakeholders signed it at a public ceremony

with the media in attendance

At first glance our impact findings appeared to show that the gang truce was an effective

mechanism for reducing violence Bivariate analyses showed a significant decline in homicides

after the truce was implemented This explained the work previously published by

67

policymakers researchers and news reporters Upon further examination of the data however

comparing change in the target and comparison areas and accounting for temporal trends we

found that the decline in homicide was part of a larger nationwide decline in violence and that

the gang truce was not responsible for the decline The only significant effect that we

uncovered was that possibly the homicides were displaced outside the target area for a brief

period of time but then returned to normal

Any one of a number of explanations might be offered for the strategyrsquos lack of effectiveness It

might be that the Jamaican gang leaders at least those in Greater August Town did not have

the organizational capacity to change gang member behavior Much prior research suggests

that in general gangs have limited organizational structure and little formal leadership This

might suggest that gangs do not possess the necessary capacity to regulate their membersrsquo

violence That said gangs in Jamaica including in Greater August Town have been found to be

fairly organizationally sophisticated and to possess strong leadership

In fact in a small number of Jamaican communities gangs have been found to be highly

organized with individual gang leaders being referred to as dons and community leaders The

gang leader in such a community is often found to have substantial control over members and

residents as these communities often turn to the don rather than the police for justice The

don will hold court and punish those who commit crime Punishment can include beatings and

torture as well as execution (Morgensen 2004) Although this level of organizational structure

and sophistication is found only in a small number of Jamaican communities generally the

gangs in Jamaica are believed to have some organizational capacity or at least enough to

reduce violence in communities

Our findings however indicated that prior to 2005 and the death of Neil Wright perhaps only

Jungle 12 could approximate that capacity to discipline members and enforce a truce After the

gangrsquos fragmentation in 2005 Jungle 12 lost much of its organizational capability and

enforcement of the truce was therefore difficult The truce negotiators sought to address the

enforcement issue by proposing a peace council that would involve all parties The proposal

was approved by all key stakeholders still some gang leaders demanded cash payments as a

68

condition for attending council meetings Peace was consistently seen by then as a bargaining

tool rather than as an honest attempt to establish and maintain peace In the end members of

only two gangs were attending the meetings21 and the council soon dissolved

In an effort to replicate the council function UWI sponsored one of the most respected

negotiators a community activist to become a one-person monitoring and intervention

specialist or a ldquoviolence interrupterrdquo His job was to ensure that truce violations did not lead to

a return of the gang wars mdash and there were many violations of the truce For example there

were instances of gang members crossing boundaries and entering the turf of another gang

armed although not initiating conflict behavior that was interpreted by the opposing gangs as

preparation for the next round of ldquowarrdquo or as laying a foundation for a surprise attack that

would exploit the truce for this purpose In the absence of the council these matters were

reported to the violence interrupter who tried to resolve the problems in consultation with the

various gang leaders Often the gang leaders were unresponsive or incapable and therefore the

threatening practices and violence continued Ultimately there were no rules or bodies or

persons who could regulate the violence and there were never any reference points for

compliance The formal truce agreement was an attempt to negotiate and impose such rules

via a collective pressure that would include third parties but it was unsuccessful in doing so

The potential for re-engineering norms related to conflict thus was not realized

Another explanation for the failure of the gang truce might be that it was more a vehicle for

rhetoric rather than for reality The gang leaders insisted that they would sign the truce

agreement only if it were ratified in public with the presence of the media (Jackson 2008 Levy

2009) The leaders might have viewed the process in and of itself as a means of increasing their

reputation and influence within the community and in policymaking circles (and to reduce

mutual distrust) In signing the truce gang leaders publicly pledged to reduce their involvement

in violence thereby calming local residentsrsquo fears They also made public efforts to increase

resources for their communities perhaps in an attempt to portray themselves as ldquoprovidersrdquo to

the community In fact the truce did provide gang leaders with an opportunity to be seen in

21 Interestingly the Jungle 12 factions did not attend any of the peace council meetings

69

public collaborating with important community stakeholders The imagery of the public signing

was of the government (via the PMI) and others approaching the gang to ask them to use their

means of informal social control in the community to reduce violence mdash to accomplish

something that the government could not do on its own As a consequence the process may

have been perceived by gang leaders as a victory because it enhanced the gangsrsquo reputation

with both the government and community

Alternatively from the start the gangs might not have been fully invested in the gang truce

One of the major criticisms of the Greater August Town gang truce was that gangs were not

required to give up their firearms although some believed that this was an unrealistic request

their demand and the demand of many that all guns be turned in immediately was

quite unrealistic given the decades of ingrained gun culture and the continued inability

of the security forces to guarantee protection for any corner against armed rivals It was

obvious to most observers that that kind of situation could not be ended overnight and

that this was a reasonable first step in the process (Levy 2009 63)

The gangs feared that if they were to disarm themselves they would be vulnerable to other

gangs and unable to protect themselves a concern that appears not to have been addressed by

mediators Indeed at times some elements within the community felt somewhat dependent on

the gangs to maintain security If the gangs would have been disarmed and there were no

near-term alternative prospects for any form of social control both the gang and the

community might have faced additional violence as has been observed in the past In the end

the gang truce only called for a reduction in gang violence and did not provide any solutions to

address the larger problems between the gangs nor did it provide the gangs with any tangible

benefits for abiding by the truce

70

Case Study C Gang Trucemdash The Honduran Experience

Introduction

Violence in Honduras is at epidemic levels increasing almost 44 percent over the past five

years In 2012 there were 7172 homicides in Honduras or about 86 homicides per 100000

population (Instituto Universitario de Democracia Paz y Seguridad 2013) making it the most

violent nation in the world (United Nations 2013) Likewise Hondurasrsquo second largest city San

Pedro Sula has the highest municipal level homicide rate in the world with 1290 homicides

(Instituto Universitario de Democracia Paz y Seguridad 2013) or about 174 per 100000

population (United Nations 2013) In comparison the average homicide rate across the globe is

about 62 per 100000 and the average homicide rate in Central America is about 27 per

100000 (United Nations 2013)22

Much of the discussion about the causes of Hondurasrsquo high homicide rate has focused on its

relationship with international drug trafficking routes gangs and conflict between crime

groups and the government and government instability Estimates of gang involvement very

widely but some have suggested that there are between 12000 (Seelke 2012) to 36000

(Ratcliffe et al 2014) gang memberrsquos in Honduras who typically belong to one of two gangs

MS-13 and 18th Street These gangs are said to be less organized than their counter parts in El

Salvador but are said to be just as involved in extortion and intimidation and perhaps more

involved in drug trafficking because of their stronger linkages with Mexican drug cartels

(Wilkinson 2013)

22 An unusual characteristic of the homicide problem in Honduras is the age of victims Typically in the Western Hemisphere homicide victims are aged 15 to 29 In Honduras however those 30 to 44 have the highest rate of violent victimization For example 1 out of 280 males 30 to 44 years old are the victim of homicide compared to 1 out of 360 males 15 to 29 years old (United Nations 2013) These findings by themselves are suggestive of a chronic gang problem (Spergel 1995) Honduras also stands out in the Western Hemisphere in the proportion of its homicides that involve a firearm In 2012 about 84 percent of the homicides involved a firearm (11) The proportion of homicides that involve a firearm appears to be increasing as well In 2008 79 of homicides involved a firearm compared to 81 in 2009 83 in 2010 (United Nations 2013)

71

Over the past decade the nation has responded with ldquoMano Durardquo (ie iron fist or heavy hand)

The new legislation provided the police with more authority to stop search and detain gang

members The new legislation also permitted the courts to sentence gang members to prison

for 12 years for simply being a member of a gang and allowed the courts to sentence

individuals to even longer prison terms for gang related incidents Concomitantly the military

joined the effort to fight gangs by patrolling neighborhoods along side the police While the

public and media strongly supported the shift in national policy toward Mano Dura much of

the evidence suggests that these legislative and policy changes were not effective as the

number of homicides continued to escalate Some suggest that its lack of success was because

gang members who were arrested were released due to of lack of evidence or those who went

to prison if they were not a gang member before entering prison joined a gang Other critics

point out that the heavy handed approach by the government led to loss in the rule of law as

vigilantes engaged in extra-judicial violence against gang members (Seelke 2012) Still others

said that the legislation and policies never really had a chance of working because of the

general lack of effectiveness of the police and courts and the wide spread corruption

throughout the criminal justice system (Zilberg 2011)

As a consequence of the above policymakers and citizens voiced optimism about the possibility

of a truce between gangs after initial results in El Salvador suggested the strategy might be

effective (Villiers-Negroponte 2013) Honduran church leaders and the Organization of

American States (OAS) began to develop a strategy to implement a similar type of truce in

Honduras and the President offered his personal support in their efforts (Arce 2013) In this

case study we examine the processes that lead to the Honduras gang truce and the nationwide

impact of the truce on homicides In the below section we discuss the major stakeholders who

participated in the truce processes leading up to the truce and the establishment of the truce

Key stakeholders

The primary facilitator for the truce process in Honduras was Archbishop Roacutemulo Emiliani who

had earlier served as the Assistant Bishop of the Dioceses of San Pedro Sula and who received

72

support from the Catholic Church to pursue the truce (Bosworth 2013) Prior to the

negotiations he was well known for his work which attempted to establish peace between the

gangs and his advocacy for prison reform and social reintegration programs for gang members

(The Daily Herald 2013) From the onset Monsignor Emiliani proceeded cautiously to ensure

reasonable expectations among the public and policymakers He maintained publically that ldquohe

didnacutet want to be a salesman of false promises about what was going to occur in the future the

things that they do are unpredictable but we expect to have a declaration of reconciliation

principles with societyrdquo (El Mundo 2013) Additionally he wanted to set reasonable

expectations because he knew that it would be a ldquoslow painful and draining processrdquo (El Nuevo

Siglo 2013) and that ldquoWhat is coming is difficult It is not easy It is complicatedrdquo (Castillo

2013)

As in El Salvador the Organization of American States (OAS) played a major role in facilitating

the peace process alongside Monsignor Emiliani Adam Blackwell served as the Secretary of

Multidimensional Security for the OAS and represented Canada on the Honduras Security

Reform Commission (Willcocks 2014) His participation in the mediation process was requested

by Honduran gang members who were in prison They requested that the OAS help broker a

peace agreement with the Honduran government and to help identify resources that would

assist gang members to obtain legitimate jobs (Associated Press 2013) The OAS together with

the Catholic Church served as a ldquobridgerdquo between the executive branch of the government and

the two gangs Additionally two of the mediators (ie Salvadoran Army officer and Police

Chaplain Monsigor Colindres and former Salvadorian congressman Mijango) who helped broker

the truce in El Salvador provided additional support to Monsignor Emiliani and Secretary

Blackwell They traveled to Honduras to present their experiences with the gang truce in El

Salvador and to convey that a gang truce is a promising and legitimate strategy for addressing

gang violence (Associated Press 2013)

Gang leaders of the two primary gangs in Honduras (MS 13 and 18th Street) also participated

extensively in the negotiation process It was stated that they had become weary of the violent

conflict and understood that a truce would be beneficial to the Honduran people (Servellon

73

2013) From the beginning however a number of the critics of the truce argued that Honduran

gangs did not have the capacity to control street level violence They characterized the

Honduran gangs as having less organizational leadership (Bossworth 2013) less control over

turf (LatinNews Daily Report 2013) and being more organizationally fractured (Dudley 2013)

than MS 13 in El Salvador

At the time that discussions about the possibility of a gang truce began President Porfirio Lobo

Sosa was publically supportive of the Catholic Church and OAS negotiating with the gangs

Media reports quoted the President saying I am ldquoprepared to do what ever is necessaryrdquo to

support the mediators (Phillips 2013) ldquoWe have to look for anything thatrsquos an alternative to

violencehellipOn the part of the government we are open to any process that can lower violencerdquo

(Associated Press 2013) and that he had given ldquohis blessing to Emillanirsquos efforts to broker peace

between the gangshelliprdquo (The Daily Herald 2013) However in November 2013 after a general

election the new president Juan Orlando Hernaacutendez through his recently appointed Vice

Minister of Security declared that the government would no longer support the truce process

with the gangs (El Heraldo 2014) Since then the Government of Honduras has not mentioned

the peace process that was initiated in May 2013

Truce making process

It is important to note that prior to the announcement of the gang truce a number of key

stakeholders were somewhat skeptical about its possibility On the one hand some suggested

that a gang truce had been attempted in the past with no success For example one

stakeholder commented to an international media outlet that ldquoEveryone here agrees itrsquos a

positive step forward but people are cautiously optimistic because in 2005 these two gangs

had another peace treaty with each other Now that treaty was very tentative it only lasted

less than two monthsrdquo (Al Jazeera 28 May 2013) On the other hand as noted above other

stakeholders believed that the local gangs did not have enough organizational leadership to

change the behavior of gang members to reduce violence (Bosworth 2013) They argued that

even if gang leaders wanted a gang truce there was no way of enforcing it on the streets

74

Several months prior to the announcement of the gang truce Carlos Mojica Lechuga an 18th

Street Salvadorian gang leader publically stated that representatives of MS13 and 18th Street

in Honduras spoke with several gang truce key stakeholders in El Salvador for the purpose of

replicating the truce in Honduras Tellingly in reflection of the visit Mojica noted that the

advantage of a gang truce is that it formally recognizes Honduran MS13 and 18th Street leaders

as important political persons within the nation He also noted that Honduran gang leaders

have historically been treated poorly and that a gang truce holds the potential for

demonstrating the political power of each of the Honduran gangs (Villiers Negroponte 2013)

The negotiators used different language to describe the early days of the truce Specifically

they mentioned that there had been a consultative process with the gang leaderships so they

were in a process like lighting23 Prior to the truce media sources mentioned that the gang

leaders were offering to stop violence and to not recruit more youth into the gangs24 The gang

leaders also spoke about the ldquopersecutionrdquo they and their family members had suffered during

the previous years highlighting that they had been prohibited social opportunities offered to

others in society (Arce 2013)

Leading up to the negotiations the leadership of both gangs expressed their interest in three

goals 1) lowering violence and crime 2) reconciliation with God society and the government

and 3) helping to improve the social conditions of their communities Although there was not

written documentation on the exact agreement between the parties one MS-13 leader

affirmed that the pact would include all violence (El Comercio 2013) However when talking

about sensitive topics such as extortions which is one of the main sources of income for the

gangs the gang leader said that ldquowould be taken up at a later daterdquo (ABC Internacional 2013)

Leaders of 18th street made similar general statements about ending violence but they were

more specific about their demands One of the 18th Street leaders stated that ldquowhat we want is

23 El Universal 2013 - httpwwweluniversalcominternacional130531obispo-hondureno-descarta-tregua-entre-las-pandillas 24 (Garcia 2013 -httpwwwlaprensahncspmediapoolsitesLaPrensaHondurasSanPedroSulastorycspcid=338546ampsid=276ampfid=98-)

75

to have a dialog with any commission appointed by President Porfirio Lobo and we are sure

that the situation in Honduras will begin to changerdquo (El Nuevo Siglo 2013)

Establishing the gang truce

On May 28th 2013 with public declarations from leaders of both gangs the gang truce was

announced From the beginning of the process the role of the government in the truce was

unclear (La Prensa 2013) Likewise there was little discussion about the exact nature of the

agreement the terms in which gang members would abide and any benefits that would be

made available to those who participated in the truce For example as one stakeholder

indicated the government never decisively considered viable proposals to give the members of

the gangs any opportunities It is important to note that none of the parties signed a formal

commitment and neither MS-13 nor 18th Street signed any type of ceasefire agreement The

gang leaders were in separate locations and were never in direct contact with each other during

the announcement That is both gangs seemingly agreed to the gang truce without ever

talking to each other The ldquopeace processrdquo was publicly announced on May 28 2013 through

ldquojoint but separaterdquo declarations made by the leaders from both gangs imprisoned in the San

Pedro Sula prison (National Penitentiary SPS) (The Daily Herald 2013) The national and

international media widely covered the declarations

The MS-13 leaders said they would not commit any more homicides or any other types of

crimes They ensured that this was an ldquoimmediaterdquo order and would be effective throughout

the country They emphasized ldquoall of the boys know what they have to do starting todayrdquo

(Pachico 2013) The leadership of 18th Street declared that they would stop violence and other

criminal activities but also indicated that the government would have to ldquolisten to themrdquo

Little research has examined whether the gang truce in Honduras ever impacted violence in the

nation Instead anecdotes have been used to portray its effectiveness One facilitator for

example indicated ldquoin Honduras the dialog with the gangs has been positive however the sad

thing in Honduras is that the two main gangs have not accepted a truce between them they

just haven`t accepted it as yetrdquo (La Prensa 2013) despite the fact that ndash in their own words ndash

76

ldquothey do want to hold a dialog with society with the government and with the policerdquo

Conversely gang leaders declared that the truce had been effective For example a leader of

18th Street noted that ldquohellipIt has already done its part telling members in the areas the gang

controls to stop the violence and crimehellip[estimating] crime had already dropped 80 percent in

those areasrdquo (Associated Press June 17 2013) Similarly a member of MS13 estimated that

violence in MS13 controlled areas declined by 45 percent (Associated Press June 17 2013) As

a symbolic gesture of the impact of the truce MS13 leaders also noted that as a gesture of good

will they made and delivered 60 beds for a nursery home in San Pedro Sula (Associated Press

June 17 2013)

Methods

For the present case study we used a pre-test post-test longitudinal quasi experimental design

Data from the 2001 Honduran Census was obtained from the National Institute of Statistics

(INE) These data provided municipal level measures of number of residents population density

per kilometer percent of population who moved in from another municipality ethnicity

percent urban number of residents immigrated to the United States percent female headed

households percent unemployed age composition income percent of households rented and

education level In addition population projections for the years 2005 2010 and 2014 were

also obtained from the National Institute of Statistics (INE) The population levels for the total

population as well as the percentage of residents in a municipality that rural and percentage of

residents who are female were linearly interpolated for the intervening years Examination of

the observed levels indicated that growth was linear overall and so we feel confident that our

linear interpolations are good approximations Second we used municipal level homicide data

by month and year for the period May 2012 through July 2014 These data were provided by

the Honduran National Police through the Honduran US Embassy Both datasets were

merged for the present analysis

Measures

77

The dependent variable for the Honduran case study is the monthly homicide rate which was

calculated by dividing the number of homicides in each municipality by its population and

multiplying this figure by 100000 We examined change by comparing the homicide data 13

months prior to the gang truce to the homicide data 14 months following the truce in each of

the nationrsquos 298 municipalities As presented in Exhibit 21 there were a total of 7910 homicides

over the study period with each municipality averaging 183 homicides (sd=837) A trend

analysis showing the monthly number of homicides on a national level prior to and following

the gang truce is presented in Exhibit 22 It shows that nationally the homicide rate gradually

declined over the study period

Exhibit 21 Summary Statistics

Pre-Truce (n=3809) Post-Truce (n=4101) Total (n=7910)

Mean SD Mean SD Mean SD

Homicide 200 900 166 773 183 837

Population 2874492 8674659 2939557 8870267 2908226 8776124

Homicide rate 521 953 413 814 465 885

Density 9926 10878 9928 10878 9927 10877

Born in the same municipality 8211 1286 8211 1286 8211 1286

Other population group 8545 2390 8545 239 8545 2390

Percent rural 7926 2732 7899 2762 7912 2748

Living in another country 331 650 331 65 331 650

Socioeconomic status 005 097 -002 102 001 100

Exhibit 22 Homicide rate in Honduras by month

78

We used census data to control for several community-level structural factors Principal

components analysis was used to reduce some of the data into a summary measure Exhibit 23

presents the results of the component loadings One component was extracted that we labeled

socio-economic status (SES) This component exhibited high loadings for percent female

headed household percent unemployment and percent completing primary education

Population was used as our exposure variable and population density (per 1 km) residential

stability percentage of the population that is rural percentage of the population that is

indigenous (verify) and the number of residents immigrated to the United States served as our

control variables Population density and outmigration were logged to address skewness in

these two measures

Exhibit 23 Factor Analysis for Socioeconomic Status

Eigenvalue 236093

Variables Factor Loadings

Female-Headed Households 07977

Unemployed 09269

Primary Education 09303

Findings

The results of our t-test analysis are shown in Exhibit 24 It shows that there was a significant

decline in the homicide rate at the municipal level before and after the gang truce The

homicide rate prior to the truce was 697 per 100000 population and following the truce the

homicide rate was 566 homicides per 100000

79

Exhibit 24 T-test on National Data by Month (n= 27 Months)

Homicide Rate Pre Truce Post Truce Change

Mean 6972346 5663592 -1308754

SE 01677361 01151668 02009224

plt 001

Next we examined the effect of the truce through ARIMA regressions The first model in Exhibit

25 does not include an effect of the temporal trend In this model we again find that homicides

declined significantly in the period following the truce We then employed the ARIMA model

again but this time included a variable (month) to control for the temporal trends in the data

Model 2 in Exhibit 25 shows that when we controlled for temporal trends the impact of the

truce we observed was no longer significant Our findings suggest that homicide did not decline

as a consequence of the gang truce but instead the decline in homicides was part of a longer

term historical trend that was independent of the truce

Exhibit 25 Results from the ARIMA Models (n= 27 Months)

Homicide Rate Model 1 Model 2

Effect SE Sig Effect SE Sig

Truce -1309 0350 -0906 0595

m2 -0623 0481

m3 -0153 1447

m4 -0074 0607

m5 -0104 0450

m6 -0050 0466

m7 -0002 0516

m8 0003 0404

m9 0024 1633

m10 -0396 0522

m11 -0754 2424

m12 0837 0366

Time -0030 0047

Intercept 6973 0254 7258 0472

Autocorrelation Coefficient 0010 0408 0111 0249

plt 001

80

Last we used a fixed effects multi-level negative binomial regression to predict the homicide

rate with population as an exposure variable and controlled for the socio-demographic

characteristics of each community The results are presented in Exhibit 26 The only significant

variable in the analysis was the impact of municipal level population stability Specifically we

found that homicide rates increased in areas where residential mobility was high Once again

our analysis showed that time had a negative effect indicating that homicides were decreasing

in all areas over the study period The gang truce itself was unrelated to the decline in

homicides

Exhibit 26 Results from the Multilevel Negative Binomial Models (n= 27 Months)

Homicide Rate Model 1 Model 2

Effect (SE) Sig Effect (SE) Sig

Fixed Effects

Homicide Rate 0004 0004

(0002)

Truce -0119 -0129

(0073) (0072)

Time -0008 -0007

(0005) (0005)

Month 1 0024 0024

(0059) (0059)

Month 2 -0107 -0107

(0062) (0062)

Month 3 -0041 -0041

(0063) (0063)

Month 4 -0033 -0033

(0065) (0065)

Month 5 -0002 -0002

(0067) (0067)

Month 6 (reference)

Month 7 0027 0027

(0051) (0051)

Month 8 0023 0023

(0057) (0057)

Month 9 0036 0036

81

(0056) (0056)

Month 10 -0051 -0051

(0057) (0057)

Month 11 -0031 -0031

(0058) (0058)

Month 12 0187 0187

(0056) (0056)

Born in the same municipality -0013

(0003)

Pecent rural -0003

(0002)

Socioeconomic status 0029

(0044)

Percent dominant population 0001

(0002)

Density (ln) -0010

(0050)

Living in another country (ln) 0001

(0061)

Intercept -10042 -10051

(0061) (0058)

ln(Alpha) -2368 -2367

(0148) (0148)

Random Effects

Var(Truce Effect) 0066 0066

(0024) (0025)

Var(Intercept) 0388 0309

(0049) (0042)

Cov(truce effect intercept) -0024 -0006

(0027) (0025)

plt 001

82

Conclusions

Violent crime in Honduras is widespread With a homicide rate of about 86 per 100000

population Honduras is one of the most violent nations in the world (United Nations 2013)

This compared to an average homicide rate around the world of 62 per 100000 people and

about 27 per 100000 in Central America (United Nations 2013) Given the high rate of violence

in Honduras it is clear that new and innovative ways of reducing violence should be considered

This case study gave an overview of the implementation and impact of the gang truce

experience in Honduras in 2013

The goal of the truce was to significantly reduce the number of homicides In order to assess

the impact of the gang truce in Honduras a pre-test post-test longitudinal quasi-experimental

design was used Census and homicide data were merged at the municipal level to asses

whether the truce had an impact on homicides controlling for population characteristics and

the natural trend in violent crime

Overall the findings suggest that while the homicide rate in Honduras was on a slight

downward trend the gang truce itself was unrelated to any homicide reductions That is the

gang truce had no measurable impact on homicides in Honduras Given the drastic reductions

achieved in El Salvador and the fact that Hondurasrsquo gang truce was a replication of El

Salvadorrsquos the following discussion will examine two important differences between the two

countries gang truces First the implementation of the truce in Honduras did not appear to

obtain trust between all parties involved and did not achieve any notable short-term

deliverables That is the implementation was not robust Second some suggest that the gangs

in Honduras do not have the organizational sophistication to be able to control their members

on the streets rendering them incapable of carrying out any truce agreements

First the implementation of the truce in Honduras did not appear robust The communication

between the parties was weak and none of the parties completed any significant actions as part

of the truce The Catholic Church and OAS served as the bridge between the two main gangs in

Honduras MS13 and 18th Street and the executive branch of the government The ldquopeace

83

processrdquo as it was called in Honduras was initiated in May 2013 After the general election in

November 2013 the newly appointed administration declared that it would no longer support

the truce process with the gangs As a result the truce was short-lived and the governmental

support for the effort shifted with the change in leadership

Though the negotiations were largely based on the good will of a well-respected negotiator

few tangible incentives were offered during the process The negotiations from the beginning

involved discussions about large scale social programming Given the short time frame these

goals in hindsight were unrealistic Naturally there was little trust between the parties at the

beginning of the negotiations and without any quick tangible deliverables from either side the

truce never really materialized

The second challenge to implementing a successful gang truce in Honduras might have been

related to the nature of the gangs themselves It was unclear whether the gangs possessed the

level of cohesion and hierarchical leadership required to make some measures feasible If the

gangs do not have the organizational capacity to control their members on the streets any

agreements that come from the negotiations would be difficult to implement In the earlier

case study of the Salvadorian experience it was suggested that the successes in that country

were related to one of the gangrsquos organizational capacity to impose control of its members The

gangs in Honduras might be different There is at least some evidence that MS 13 in Honduras

might have less organizational leadership (Bossworth 2013) less control over turf (Latin News

Daily Report 2013) and have been more organizationally fractured (Dudley 2013) than their

counterparts in El Salvador The gangrsquos ability to operate as an efficient organization can greatly

impact the outcome of the truce process In general we know that gangs do not have very high

levels of organizational sophistication (Decker Katz and Webb 2008 Decker Bynum Weisel

1998) It might be that gangs in Honduras are more the norm in terms of organizational

capacity

In summary the 2012 truce negotiations in Honduras did not produce any measurable

reductions in the homicide rate The gangs wanted to speak to Honduran society and they even

84

preferred to speak with the governmental authorities but they never interacted with either

The negotiation process seemed to end as quickly as it started The gangs did not deliver with

lower rates of violence and the government did not provide social programs It would be safe to

say that a robust gang truce did not materialize in Honduras The implementation of the truce

seemed to struggle for two primary reasons First the mediators were not able to accomplish

any quick wins to build trust between the parties involved The commitment level on all sides

was not clear throughout the process Second it is not clear whether the gangs in Honduras

have the organizational capacity to control members on the street as would be required to

carry out an effective gang truce In short the 2013 gang truce in Honduras was unsuccessful

85

Conclusions Policy Implications and Recommendations

The purpose of this report was to systematically examine the effectiveness of gang truces Gang

truces have been widely implemented but rarely evaluated Of those gang truces that have

been evaluated little attention has been given to why and how they came into existence In

this report we reviewed prior research on gang truces and presented case studies of gang

truces implemented in El Salvador Jamaica and Honduras for the purpose of understanding

the negotiation processes undertaken with and between gangs and other stakeholders We

were interested in identifying the actors involved in the negotiations the goal(s) of the

negotiations and the strategies employed to carry them out Most importantly however we

wanted to determine whether the gang truce resulted in a reduction in the number of

homicides Each case study offers lessons learned that are unique to their particular

circumstances and when considered together provide direction to policymakers on the

benefits and risks of implementing gang truces

The case studies presented here constitute the most comprehensive evaluations of gang truces

to date Existing documents were used to collect information about the processes associated

with each gang truce Many of these documents included such items as peer reviewed articles

books and reports The majority of these documents were collected over the Internet

requesting documents from those close to the truce and searching library databases Related

the case studies made use of articles obtained from local newspapers The newspaper articles

were not only intended to provide a historical record of the development of each gang truce

but also to provide additional insight into the various external forces that might have impacted

the gang negotiations Because the newspaper serves as a forum for the community to speak

about its concerns newspaper articles also provided a rich source of data on how those in the

community felt about the gang truce Accordingly the newspaper articles offered a different

view of the problem and offered different opinions as to how a gang truce should or should not

be implemented We also conducted a small number of in-depth qualitative interviews with

key informants These data were collected to supplement existing documents and to clarify

issues associated with the negotiation processes This included but was not limited to

86

questions pertaining to identifying the actors involved in the negotiations the goal(s) of the

negotiations and strategies employed to carry out negotiations The interviews were intended

to obtain information from those who possessed first hand knowledge about the gang truce in

each nation

We examined the impact of each truce using official data We first performed a simple t-test

comparing the homicide rates before and after the truce However as discussed above this

technique has limitations The most severe of which is that even if the test was significant it

would be difficult to determine whether the difference was due to the gang truce or a natural

change over time in the outcome We addressed this limitation by using time series models

whereby the homicide rate for the community was modeled as a function of time with truce

period indicators included to measure the effect of the truce net of the temporal trends These

models were estimated with ARIMA techniques Supplemental models were also employed to

examine and control for factors other than the truce that might have impacted homicide over

the study period

Summary of Findings Related to the Implementation of a Gang Truce in the Three Sites

We found that the implementation of gang truces have a number of common characteristics

The first is that in each case a community was experiencing an uncharacteristically high number

of gang related homicides over a fairly lengthy period of time The continued high level of

violence in each case resulted in the community placing strong pressure on the government in

general and the justice system in particular to respond to the problem quickly and effectively

In each case they had first attempted to control gang violence through suppression oriented

strategies and these strategies were found to be ineffective over the intermediate and long

term In turn each communityrsquos inability to exercise traditional informal and formal social

control to decrease levels of violence became self evident to the public and government This

resulted in both the state and community to seek (or participate in) an alternative strategy in

which negotiators would formally andor informally work with gang leaders to establish a truce

that would reduce gang homicide

87

Key stakeholders involved in the negotiation and establishment of each gang truce were fairly

similar In each of the cases examined the gang leaders of the largest and most violently

involved gangs were willing to consider participating in negotiations that could lead to a truce

In each of these cases it was clear that the gangs not only sought to collaborate with the

negotiators for the purpose of reducing violence but perhaps more importantly were seeking a

means in which to gain greater more positive recognition in the community and to reap some

form of benefit to themselves their members and possibly their community In each case

while not always formally involved government officials were at a minimum made aware of

negotiations and in some cases solicited the assistance of third partyrsquos to broker an agreement

between stakeholders In each case it was at least implicitly understood that the government

would ldquolistenrdquo to the gang leaderrsquos expectations of the government and what theymdashthe gang

leaders--had to offer in exchange We found that when the government was no longer willing

to ldquolistenrdquo to or collaborate with negotiators the truce processes ended abruptly Negotiators

were typically comprised of a very small group (ie 2-3) of individuals who were perceived to

be ldquohonest brokersrdquo In El Salvador and Honduras this included a high ranking Catholic Church

official a leader from an international diplomatic organization (ie OAS) and other neutral

parties In Jamaica this included a quasi-governmental organization that had been established

for the purpose of brokering negotiations between gangs to reduce violence and the local

university which had access to staff who were perceived to be neutral but had an interest in

reducing violence due to its proximity to the university

The strategies used to execute each gang truce were similar but yet importantly different They

were similar in that each involved a team of negotiators working to identify common goals to

be achieved and identifying tangibles that could be delivered to the gang leaders gang

members and their community in exchange for the gang achieving their stated goals They

were different however in terms of the structure of the delivery of each parties promise to the

other In Honduras and Jamaica it appears that gang leaders committed to reducing gang

violence in exchange for general promises made by the negotiators for example that

substantial public works programs would be implemented for the goal of reducing

unemployment among gang members and the community In both of these cases it required

88

the government to develop and deploy large scale social programming in a very quick period of

timemdashsomething that neither government had a strong record of demonstrating In El Salvador

negotiators employed a strategy of the gang leaders promising to deliver immediate changes in

gang member behavior for immediate administratively natured changes by the government

For example in exchange for a reduction in gang violence the government agreed to relocate

imprisoned gang leaders to less restrictive prisons and provide them some privileges Following

the successful execution of the first part of the truce which resulted in near term success for

both parties they began to negotiate issues that would take a longer period of time for the

gangs and government to deliver Our findings suggested that some promised deliverables need

to be easily and quickly delivered early in the process so that trust increases between both

parties Stakeholders only have a brief period of time to provide promised benefits before trust

is lost and that tangible benefits need to be delivered in weeks or months not years

Summary of Findings Related to the Impact of Gang Truces in the Three Sites

Our findings suggest that El Salvadorrsquos gang truce had a significant and dramatic impact on the

homicide rate The mean number of monthly homicides declined from about 354 prior to the

truce to about 218 following the truce for a net decrease of about 136 homicides per month

Our forecast showed that between March 2012 and June 2014 the truce had saved about 5501

lives However there was no significant change between the pre-truce and post-truce periods

in the number of thefts extortions robberies rapes and auto theftsrobberies We also found

that the gang truce did not result in a homogenous decline in violence across municipalities

About 61 percent of municipalities experienced a decline in homicides but the decline in

violence varied substantially between municipalities We examined this issue further by parsing

out the relative influence of the number of MS13 and 18th Street gang members on the street

and in prison from each municipality Our analyses indicated that following the truce the

number of MS13 and 18th Street gang members on the street in a municipality was not

significantly related to a decline in homicide but the number of imprisoned MS13 and 18th

Street gang members was associated with a significant change in homicides following the gang

truce In particular the number of imprisoned MS13 gang members was associated with a

89

significant decline in homicides following the gang truce and the number of imprisoned 18th

Street members was associated with a significant increase in homicides following the truce

In Jamaica our initial findings showed that the gang truce might be an effective mechanism for

reducing violence Bivariate analyses showed a significant decline in homicides immediately

after the truce was implemented This explains the work previously published by policymakers

researchers and news reporters Upon further examination of the data however comparing

change in the target and comparison areas and accounting for temporal trends we found that

the decline in homicide was part of a larger nationwide decline in violence and that the gang

truce was not responsible for the decline The only significant effect that we uncovered was the

possibility that homicides were displaced outside the target area for a brief period of time but

then returned to normal

Our findings from Honduras told a similar story as Jamaica Initial analysis showed that the

number of homicides on average declined across municipalities following the gang truce

Specifically the mean number of homicides declined by 13 per 100000 population with 687

homicides per 100000 population on average occurring in each municipality prior to the truce

and 566 homicides per 100000 population on average occurring in each municipality after

the truce However after we examined the effect of the truce through the ARIMA model and

included a variable (month) to control for the temporal trends in the data the impact of the

truce we observed in our bivariate analysis was no longer significant Our findings as in

Jamaica suggested that the decline in homicides was not the consequence of the gang truce

but instead the decline in homicides was part of a long term decline in homicides due to

exogenous factors

The Potential Benefits and Consequences of a Gang Truce

Over the last several years there have been a number of naturally occurring experiments

involving gang truces in a variety of nations in various regions of the world Findings from

evaluations of gang truces are mixed As noted above in El Salvador the gang truce could be

characterized as highly effective at least for the two years following the truce It is worth

mentioning that even after the truce breakup homicides rates while above truce levels

90

continue slightly below pre-truce levels In Jamaica and Honduras the gang truce had no impact

on violence In Los Angeles and Trinidad there was evidence that violence decreased for at

least ninety days but then increased substantially beyond those rates observed prior to the

gang truce (see the introduction section of this report for this discussion) As a consequence it

appears that the potential for long term consequences might out weigh the potential for short

term benefits Only one study site(El Salvador) demonstrated a truce having a substantial and

long term impact on violence Others conversely demonstrated the truce had no impact or

increased violence over the long term In fact a number of scholars have noted that gang truces

are likely to result in a boomerang effect with gang violence increasing over the long run

because of enhanced cohesion within the gang (Klein 1995) Maguire (2013) notes that when

government officials negotiate a truce with gangs they might ldquoinadvertently be acknowledging

gangs as legitimate social entitiesrdquo (p 11) This in itself might increase cohesion among gangs

which has been found to be associated with increased levels of criminality (Decker et al 2008

Klein 1971 Maguire 2013)

Further research is needed to examine how gang truces might impact group cohesion and if it

does whether this in turn results in greater violence Gang truces convey the well-intentioned

image that violence has been addressed and policymakers are doing something about the

problem but researchers need to better understand the probability of a gang truce reducing

violence increasing violence or having no impact This will better position policymakers to

understand the relative risks associated with these types of interventions

Our findings also suggest that while gang truces could be an effective intervention in areas

where gangs are highly structured and organized such as El Salvador they could be counter-

productive in areas where gangs are not as structured and organized Because the vast majority

of street gangs are not well organized (Klein 1995 Spergel 1995) the utility of a gang truce in

reducing violence might be limited Our findings coupled with prior research suggest that gang

interventions need to be tailored to the nature of the gang and its members or it risks

increasing gang violence

Final Thoughts

91

Our analysis suggests that gang truces should only be used as a means of last resort and then

only under certain conditions Given the risks associated with a gang truce communities with

high levels or at least modest levels of formal social control should rely on other more

promising gang control strategies Only when the state has limited or greatly reduced capacity

for social control should a truce be considered Concomitantly a gang truce should only be

considered when a community is experiencing a substantial amounts of gang violence

Communities that are experiencing minimal to modest amount of gang violence may have more

to lose from the establishment of a gang truce than they have to gain Additionally our

findings suggest that a gang truce might only be feasible when gangs are sufficiently organized

to the extent that they have the capacity to regulate memberrsquos behavior In other words gang

leaders must have the ability to reduce their memberrsquos involvement in violence for a gang truce

to work Our findings more concisely suggest that gang truces should only be considered when

there are a great number of gang homicides the state has limited capacity to address the

problem and gang leaders have enough informal social control over their members that they

themselves can substantially control the levels of violence in their community by regulating

their memberrsquos behavior

Recommendations

Gang truces are conjunctural strategies States who suffer from gang-related violence must

establish permanent public policies for crime control and prevention A government that

considers implementing a gang truce should be aware that it cannot become the center

strategy of its public policy for citizen safety

Gang truces should only be used as a means of last resort and then only under certain

conditions Stakeholders must determine whether a process of dialog or negotiation with gangs

is legal ethical and feasible

Stakeholders must anticipate demands that are likely to arise and their response options Some

demands may be easily met such as improved prison conditions Others are much more

difficult and amorphous such as community development through more integrated violence

92

prevention programs (such as those implemented by SolucionES in El Salvador) local economic

development programs or economic reinsertion of ex-gang members

Stakeholders should incorporate immediately achievable and demonstrable deliverables Long-

term goals and promises are unlikely to create the trust needed to sustain a gang truce

Stakeholders must first determine the position in which they are negotiating the incentives that

are possible to deliver and the boundaries and limits they face Gangs are mostly likely to trust

representatives from NGOs community-based organizations and members of the faith based

community as brokers because they are considered more reliably neutral advocates for peace

They need to understand the capacity of the government to deliver promises in a timely

manner

Governments have to make a choice about the visibility and transparency of its participation

This decision needs to be made in the context of the national and local laws the publicrsquos

expectations of transparency and patterns of practices of the past

Governments must be strategic in their support for truce initiatives Some donor funded

programs run by the government prohibit gang member participation and if the government

does not receive approval from the donor it may risk the donor withdrawing its sponsorship of

the program

Governments must ensure an inter-institutional coordination for the management of truces to

avoid the responsibility to be of a single government institution It is necessary to generate or

collect reliable and pertinent data that can be used to analyze and assess the process

It is necessary to implement an effective monitoring system of the truce process similar to that

used in the full report as it can help parties understand what is working and who is delivering

on their promises More specifically the monitoring and truce management system should be

able to identify truce violations and be prepared to respond through the use of legal and

effective practices if stakeholders do not comply

93

Finally it is necessary to develop evaluations of gang truces and monitoring programs and

support violence prevention activities local economic development activities and pilot

programs to support the reinsertion of ex-gang members into society Clearly national

governments municipal governments NGOs and community-based organizations need

increased capacity and resources to discourage the growth of gangs among at-risk youth It is

therefore increasingly important to create economic opportunities for gang members willing to

leave the gangs and find other legal employment Developing and sustaining those

opportunities in nations with high incidences of poverty will require significant international

funding

94

References

Anderson Elijah 1999 Code of the Street New York Norton

mdash 1998 ldquoThe Social Ecology of Youth Violencerdquo In Crime and Justice vol 24 edited by M

Tonry 65-104

mdash 1990 Street Wise Chicago University of Chicago Press

Arce A (2013 May 28) Honduras gangs declare truce ask talks with govt Associated Press

The Big Story Retrieved from

httpbigstoryaporgarticlehonduras-gangs-declare-truce-seek-talk-govt

Archibold Randal 2012 ldquoGangsrsquo Truce Buys El Salvador a Tenuous Peacerdquo The New York Times

Originally published online on August 27 2012

Axelrod R (1986) ldquoAn Evolutionary Approach to Normsrdquo American political science review

80(04) 1095-1111

Ayala Edgard 2012 Gangs Back Plan for Violence-Free Districts in El Salvador

httpwwwipsnewsnet201212gangs-back-plan-for-violence-free-districts-in-el-

salvador Found on November 4 2014

Bakrania Shivit 2013 Policy Responses to Criminal Violence in Latin America and the

Caribbean GSDRC Helpdesk Research Report 934 Birmingham UK GSDRC University

of Birmingham

Bargent James 2013 ldquoRise in Disappearances Feeds Doubts Over El Salvador Trucerdquo August 1

2013 Found at httpwwwinsightcrimeorgnews-briefsrise-in-disappearances-in-el-

salvador-feeds-doubts-over-truce on September 1 2014

Battin Sara R Karl G Hill Robert D Abbot Richard F Catalano and J David Hawkins 1998

ldquoThe Contribution of Gang Membership to Delinquency Beyond Delinquent Friendsrdquo

Criminology 36 93-115

Block Richard 2000 ldquoGang Activity and Overall Levels Of Crimerdquo Journal of Quantitative

Criminology 16 (3) 369-83

ii

have been studied to various extents including those in Los Angeles and Trinidad and Tobago

Important differences in how the various truces were negotiated may explain the different

results and one important difference ndash the ability of government and non-gang community

stakeholders to promise and immediately produce measurable deliverables ndash appear to be

especially significant

Prior Evaluations of Gang Truces

Little research to date has examined the effectiveness of gang truces and much of the research

that does exist has been restricted to the field of public health Studies of a gang truce in South

Central Los Angeles between the Crips and Bloods by Cotton (1992) and Ordog et al (1993

1995) found temporary reductions in the number of homicides and gunshot wounds during the

truce but these studies failed to point out that despite about a 35 percent decrease in

homicides for the first three months of the truce

homicides then doubled in months four through

eleven compared with the pre-truce period

Similar findings were reported in Trinidad and

Tobago (Maguire Katz and Wilson 2013) where

violence declined for a brief period of time

(again for about three months) but then

increased substantially over the long term (12

months) These results suggest that gang truces

may produce short-term benefits yet result in

long-term adverse consequences

Additionally these studies have suffered from a

variety of flaws including 1) poor theoretical

assumptions about gangs including often

ignoring that they might have the organizational

structure and internal discipline to enforce their

truces 2) relatively weak evaluation designs that fail to account for other potential causes for

What is a Gang Truce

A gang truce is a nonviolent resolution to a larger conflict between groups that has an impact on general levels of violence and other forms of criminality within a community (Ordog et al 1993 1995 Whitehill et al 2012) It differs from conflict interruption resolution or mediation efforts which seek to rapidly intervene typically through outreach workers or violence interruption specialists in episodic violent events between groups in a community Gang truces often involve dialogue and negotiations between multiple parties (eg gangs government NGOrsquos religious organizations) that seek to recalibrate the norms of conflict within and between groups for the purpose of reducing or eliminating violence and other crime

iii

or broader trends related to violence reductions 3) a failure to examine the processes involved

in creating gang truces 4) a failure to examine unintended consequences of the truces beyond

their impact on gun violence and 5) a lack of sophisticated statistical analysis capable of

discerning overall trends cyclic patterns outliers and turning points

Overview of Study Design

This policy brief examines gang truces implemented in El Salvador Jamaica and Honduras

including the processes undertaken with and between gangs and other stakeholders We

collected information about the processes associated with each gang truce from a variety of

sources including peer reviewed articles books reports local newspaper articles and

interviews of persons with first-hand knowledge about the gang truce in each nation We

examined the impact of each truce using official data We first performed a simple t-test

comparing the homicide rates before and after the truce We then employed time series

models not used in other studies to evaluate homicide rates as a function of time with truce

period indicators included to measure the effect of the truce net of the temporal trends We

also employed supplemental models to examine and control for factors other than the truce

that might have affected homicides over the study period

Research Findings

Implementing a Gang Truce

The gang truces studied have a number of common characteristics First in each case a

community was experiencing an uncharacteristically high number of gang related homicides

over a fairly lengthy period of time which resulted in each community placing strong pressure

on the government in general and the justice system in particular to respond to the problem

quickly and effectively Second each community had first attempted unsuccessfully to control

gang violence through suppression-oriented strategies Third each communityrsquos inability to

exercise traditional informal and formal social control to decrease levels of violence became

self-evident to the public and government This resulted in all involved stakeholders wanting to

seek (or participate in) an alternative strategy in which brokers would formally andor

informally work directly with gang leaders to establish a truce that would reduce homicides

iv

Key stakeholders involved in the negotiation and establishment of each gang truce were also

somewhat similar In each case examined the leaders of the largest and most violently

involved gangs were willing to participate In each of these cases it was clear that the gangs

sought to collaborate with the brokers not only for the purpose of reducing violence but

perhaps more importantly as a means to gain greater more positive recognition in the

community and to reap some form of benefit to themselves their members and perhaps their

community In each case while not always formally involved government officials were at a

minimum made aware of negotiations and in some cases solicited the assistance of third

parties to broker an agreement between stakeholders In each case it was at least implicitly

understood that the government would ldquolistenrdquo to the gang leaderrsquos expectations and offers

We found that when the government was no longer willing to ldquolistenrdquo to or collaborate with

brokers the truce processes ended abruptly Brokers were typically comprised of a very small

group (ie 2-3) of individuals who were perceived to be ldquohonest brokersrdquo In El Salvador and

Honduras this included a high ranking Catholic Church official a leader from an international

diplomatic organization (ie OAS) and other neutral parties In Jamaica this included a quasi-

governmental organization that had been established for the purpose of brokering negotiations

between gangs for the purpose of reducing violence as well as the local university which had

access to staff who were perceived to be neutral but had an interest in reducing violence near

the university

The strategies used to execute each gang truce were generally similar but had important

differences Each involved a team of brokers working to identify common goals to be achieved

and tangibles that could be delivered to the gang in exchange for the gang achieving the stated

goals They differed by how each partyrsquos promise was delivered In Honduras and Jamaica gang

leaders traded violence reduction pledges for long-term government and social changes such

as the development of substantial public works programs to reduce unemployment In

Honduras and Jamaica the government was asked to develop and deploy large scale social

programming in a short time framemdashsomething for which neither country had a strong track

record In El Salvador by contrast brokers secured promises for immediate changes in gang

behavior in exchange for feasible immediate deliverables from the government For example in

v

exchange for a reduction in gang violence the government agreed to immediately relocate

imprisoned gang leaders to less restrictive prisons and provide them some privileges Following

the successful execution of the first part of the Salvadoran truce which resulted in near term

success for both parties they began to negotiate broader issues that would take longer for both

sides to deliver Our findings suggest that some promised deliverables need to be easily and

quickly delivered by both parties early in the process to achieve trust and serve as a first test

of gang leadersrsquo ability to deliver Stakeholders have only a brief period of time to provide

promised benefits before trust is lost meaning that tangible benefits need to be delivered in

weeks or months not years

The Impact of a Gang Truce

El Salvador We found that El

Salvadorrsquos gang truce had a definite

impact on the homicide rate The

mean number of monthly homicides

declined from about 354 prior to the

truce to about 218 following the

truce for a net decrease of about

136 homicides per month Our data

show that between March 2012 and

June 2014 the truce had saved

about 5501 lives (see exhibit 1)

From a hypothetical stand it is

possible to make the assumption that a number of these deaths averted could have been

transformed in disappearances and therefore they were not counted within the official

homicide statistics However the results from the analysis point out that the number of

disappearances was not significantly related with the change in the global behavior of

homicides Additionally over the period of analysis (January 2010 to June 2014) there was no

significant change between the pre-truce and post-truce periods in the number of thefts

extortions robberies rapes and auto theftsrobberies

Exhibit 1 Forecast of homicides without gang truce

vi

We also found that the gang truce did not result in a homogenous decline in violence across

municipalities About 61 percent of municipalities experienced a decline in homicides but the

decline in violence varied substantially between municipalities We studied this issue further by

examining the impact of the initiative ldquoFree Violence Municipalitiesrdquo and it was found that the

behavior of violence in those municipalities was not significantly linked to the initiative but

rather to the general dynamic of the truce process nationwide Additionally we parsed out the

relative influence of the number of MS13 and 18th Street gang members on the street and in

prison from each municipality Our analyses indicated that following the truce the number of

MS13 and 18th Street gang members on the street in a municipality was not significantly

related to a decline in homicide but the number of imprisoned MS13 and 18th Street gang

members from the municipality was In particular the number of imprisoned MS13 gang

members from a municipality was associated with a significant decline in homicides in that

municipality following the gang truce and the number of imprisoned 18th Street members

from a municipality was associated with a significant increase in homicides in that

municipality following the truce These findings lend support to the idea that MS13 is more

organized than the typical street gang and that imprisoned MS13 gang members exhibit strong

influence over their fellow gang members on the street Our findings also suggest however

that the gang truce had a boomerang effect in municipalities with high numbers of imprisoned

18th Street members implying that 18th Street might not have as much organizational capacity

to regulate violence on the streets as MS13 The truce provided incarcerated MS13 and 18th

Street gang leaders an opportunity to negotiate with high-ranking officials and influential

diplomats including representatives of the Organization of American States This may have

increased their legitimacy inside and outside of their gangs It appears that MS13 was able to

exert its span of control over the communities in which they had influence and they were able

to deliver on the terms of the gang truce negotiations In the case of 18th Street however

incarcerated gang members may not have had the same organizational capacity for

communicating and carrying out directives In fact a review of the gang truce indicated that

there was a conflict taking place between two factions within 18th Street Consequently the

organizational structure and culture of 18th Street might be more diffuse than that of MS13

vii

and its leadership structure might not be as strong because of the internal fractures within the

gang This might further explain why homicides increased in 18th Street communities The

internal fractures within the 18th Street gang may have resulted in intra-gang violent conflict

that was largely contained within 18th Street controlled territories

Jamaica In Jamaica at first glance our impact findings appeared to show that the gang truce

might be an effective mechanism for reducing violence Bivariate analyses showed a significant

decline in homicides immediately after the truce was implemented This explains the work

previously published by policymakers researchers and news reporters Upon closer

examination of the data however comparing change in the target area to other areas in

Jamaica and accounting for temporal trends we found that the decline in homicide was part of

a larger nationwide decline in violence and that the gang truce was not responsible for the

decline The only significant effect that we uncovered was the possibility that homicides were

displaced outside the target area for a brief period of time but then returned to previous

levels

Honduras Our impact findings from our analysis of data from Honduras tell a very similar story

as Jamaica Initial analysis showed that the number of homicides on average declined across

municipalities following the gang truce Specifically the mean number of homicides declined by

12 per 100000 population from an average of 687 per 100000 population in each

municipality before the truce to an average of 566 thereafter However after we examined the

effect of the truce through time series analysis and included a variable (month) to control for

the temporal trends in the data the impact of the truce we observed in our bivariate analysis

was no longer significant Our findings as in Jamaica suggest that the decline in homicides was

less a consequence of the gang truce than of a broader short to medium term trend And

importantly in both Jamaica and Honduras the respective governments failed to deliver on

gangsrsquo demands for large-scale social and employment programs

Conclusions and Policy Implications

Over the last several years there have been a number of naturally occurring experiments

involving gang truces in a variety of nations and regions Findings from evaluations of gang

viii

truces are mixed In El Salvador the gang truce could be characterized as an effective short

term strategy to reducing homicides It is worth mentioning that despite homicides rates are

above truce levels they continue slightly below pre-truce levels In Jamaica and Honduras the

gang truce had no short or medium-term impact on overall violence In Los Angeles and

Trinidad y Tobago there was evidence that violence decreased for at least ninety days but then

increased substantially beyond those rates observed prior to the gang truce Thus the

effectiveness of the truce in El Salvador appears to be isolated and must be evaluated within

the context of other truces that have failed to reduce homicide violence Policy makers must

evaluate whether the conditions that allowed short term effectiveness of the gang truce in El

Salvador (such as the ability to promise and deliver immediate results) exist in other violent

areas before evaluating whether a truce strategy might be appropriate And they should be

heavily cautioned that the potential for long term negative consequences might outweigh the

potential for short term benefits

Indeed it is important to note that a number of scholars have noted that gang truces are likely

to result in a boomerang effect with gang violence increasing over the long run because of

enhanced cohesion within the gang (Klein 1995) Maguire (2013) notes that when government

officials negotiate a truce with gangs they might ldquoinadvertently be acknowledging gangs as

legitimate social entitiesrdquo (p 11) This in itself might increase cohesion among gangs which has

been found to be associated with increased levels of criminality (Decker et al 2008 Klein 1971

Maguire 2013) Hence it is important to consider the fact that gangs are illegal groups in El

Salvador and it should be cautious when carrying out dialog or negotiation processes with

them Further research is needed to examine how gang truces might impact group cohesion

and if it does whether the cohesion created could be effectively directed toward more

productive non-violent endeavors Gang truces convey the well-intentioned image that violence

has been addressed and policymakers are doing something about the problem but unless the

truce is implemented in a manner and under conditions where immediately achievable results

can be promised delivered and measured there remains a significant chance that the truce will

fail or worse yet backfire Thus it is imperative that any type of concession made by

ix

Governments to gangs within a truce framework should be transparent so that all sectors of

society have certainty that every action is being done within the existing rule of law2

In the case of El Salvador the truce arises from to the absence of effective public policies and

practices for violence control and prevention The truce was planned as a strategy to reduce

gang-related homicides During the process different organizations got involved including

religious (facilitating and protecting human rights) non-government (managing and facilitating

dialog and negotiation processes) international (providing funding for insertion programs) and

government (facilitating and providing certain conditions for dialog and negotiation)

organizations Some of the concessions that the Government provided in order to achieve a

reduction of homicides were within the law but others generated confusion and they seemed

to be close to the legal or socially acceptable limits This fact along with the poor transparency

of authorities in the management of the practice with public media and public weakened the

process and postponed its continuity

The present study suggests that gang truces should only be used as a means of last resort

and then only under certain conditions Given the risks associated with a gang truce

communities with high levels or at least modest levels of formal social control should rely on

other more promising gang control strategies such as pulling levers (ie Boston Ceasefire)

community oriented policing and the Gang Resistance Education and Training (aka GREAT)

program Only when the state has limited or greatly reduced capacity for social control should a

truce be considered Concomitantly a gang truce should be considered as an alternative only

when a community is experiencing a substantial amount of gang violence Communities that

are experiencing minimal to modest amount of gang violence may risk more from the

establishment of a gang truce than they have to gain Additionally our findings suggest that a

gang truce might only be feasible when gangs are sufficiently well organized to be able to

regulate their membersrsquo behavior and cause their members to behave less violently In El

22 Pragmatically this issue is complicated On the one hand transparency is a foundational element within a democracy and is necessary to ensure proper oversight of the government On the other hand it might not be possible to implement a gang truce with too many actors having a voice Policymakers might consider creating a policy that allows such negotiations take place but requires particular actors (such as a judicial body) to be informed of the process to ensure transparency and adherence to the rule of law

x

Salvador there is evidence of the strong organizational structure of gang MS13 Among other

factors the magnitude of its membership the chain of command from its leaders in prison and

the discipline of its leaders in the streets seemed confirmed On the contrary the

organizational structure of gang Barrio 18 ndash divided in two factions fighting over the leadership

ndash showed to be a less stable counterpart within the truce

Finally dialog and negotiations processes with or between gangs must have the capacity to

promise and deliver immediate benefits to the gangs that gang members can see or experience

in order to secure their continued participation in the truce as well as the capacity to monitor

and respond to truce violations Most importantly any effort aimed at reducing violence is

important and should be examined and assessed but it must have a transparent foundation

especially when it affects population rights as a whole

General recommendations

Gang truces are conjunctural strategies States who suffer from gang-related violence must

establish permanent public policies for crime control and prevention A government that

considers implementing a gang truce should be aware that it cannot become the center

strategy of its public policy for citizen safety

Gang truces should only be used as a means of last resort and then only under certain

conditions Stakeholders must determine whether a process of dialog or negotiation with gangs

is legal ethical and feasible

Stakeholders must anticipate demands that are likely to arise and their response options Some

demands may be easily met such as improved prison conditions Others are much more

difficult and amorphous such as community development through more integrated violence

prevention programs (such as those implemented by SolucionES in El Salvador) local economic

development programs or economic reinsertion of ex-gang members

Stakeholders should incorporate immediately achievable and demonstrable deliverables Long-

term goals and promises are unlikely to create the trust needed to sustain a gang truce

Stakeholders must first determine the position in which they are negotiating the incentives that

are possible to deliver and the boundaries and limits they face Gangs are mostly likely to trust

xi

representatives from NGOs community-based organizations and members of the faith based

community as brokers because they are considered more reliably neutral advocates for peace

They need to understand the capacity of the government to deliver promises in a timely

manner

Governments have to make a choice about the visibility and transparency of its participation

This decision needs to be made in the context of the national and local laws the publicrsquos

expectations of transparency and patterns of practices of the past

Governments must be strategic in their support for truce initiatives Some donor funded

programs run by the government prohibit gang member participation and if the government

does not receive approval from the donor it may risk the donor withdrawing its sponsorship of

the program

Governments must ensure an inter-institutional coordination for the management of truces to

avoid the responsibility to be of a single government institution It is necessary to generate or

collect reliable and pertinent data that can be used to analyze and assess the process

It is necessary to implement an effective monitoring system of the truce process similar to that

used in the full report as it can help parties understand what is working and who is delivering

on their promises More specifically the monitoring and truce management system should be

able to identify truce violations and be prepared to respond through the use of legal and

effective practices if stakeholders do not comply

Finally it is necessary to develop evaluations of gang truces and monitoring programs and

support violence prevention activities local economic development activities and pilot

programs to support the reinsertion of ex-gang members into society Clearly national

governments municipal governments NGOs and community-based organizations need

increased capacity and resources to discourage the growth of gangs among at-risk youth It is

therefore increasingly important to create economic opportunities for gang members willing to

leave the gangs and find other legal employment Developing and sustaining those

opportunities in nations with high incidences of poverty will require significant international

funding

xii

References

Cotton Paul 1992 ldquoViolence Decreases with Gang Trucerdquo JAMA 268(4) 443-44

Decker Scott H 2003 Policing Gangs and Youth Violence Belmont CA Wadsworth

Decker Scott H and W Reed 2002 Responding to Gangs Evaluation and Research

Washington DC National Institute of Justice

Decker Scott H Charles M Katz and Vincent J Webb 2008 ldquoUnderstanding the Black Box of

Gang Organization Implications for Involvement in Violent Crime Drug Sales and

Violent Victimizationrdquo Crime and Delinquency 54 153-72

Esbensen Finn-Aage and D W Osgood 1997 ldquoNational Evaluation of GREATrdquo US

Department of Justice Office of Justice Programs National Institute of Justice

Katz Charles M and Webb Vincent J 2006 Policing Gangs in America New York Cambridge

University Press

Klein Malcolm W 1995 The American Street Gang New York Oxford University Press

mdash 1971 Street Gangs and Street Workers Englewood Cliffs NJ Prentice-Hall

Maguire Edward Charles Katz and David Wilson 2013 ldquoThe Effects of a Gang Truce on Gang

Violencerdquo Unpublished paper Washington DC American University

Maguire Edward 2013 Research Theory and Speculation on Gang Truces Woodrow Wilson

International Center for Scholars in Washington DC

McCorkle R and T Miethe 2002 Panic The Social Construction of the Street Gang Problem

Upper Saddle River NJ Prentice-Hall

Ordog Gary J W Shoemaker Jonathan Wasserberger and Michael Bishop 1995 ldquoGunshot

Wounds Seen at a County Hospital Before and After a Riot and Gang Truce Part Twordquo

Journal of Trauma-Injury Infection amp Critical Care 38(3) 417-19

Ordog Gary J Jonathan Wasserberger Julius Ibanez Michael Bishop Eduardo Velayos

Subramaniam Balasubramanium and William Shoemaker 1993 ldquoIncidence of Gunshot

Wounds at a County Hospital Following the Los Angeles Riot and a Gang Trucerdquo Journal

of Trauma 34 779-82

Whitehill J M Webster D W amp Vernick J S 2012 ldquoStreet conflict mediation to prevent

youth violence Conflict characteristics and outcomesrdquo Injury prevention injuryprev-

2012

1

Introduction

Given the increasingly devastating individual family and community effects of gang activity

over the past several decades an increasing body of literature has focused on gangs gang

members and gang activity A core theme running throughout this body of literature is that

gang members are significantly more likely to be the offenders and victims of violent crime than

non-gang members (Battin Hill Abbott Catalano and Hawkins 1998 Esbensen Winfree He

and Taylor 2001 Katz et al 2011 Huff 1998 Curry et al 2002 Miller and Brunson 2000

Miller and Decker 2001 Pyrooz et al 2012 Curry et al 2001) and disproportionately affect

neighborhood levels of crime and violence (Block 2000) These findings have been robust in

that they have been repeatedly found regardless of research methodology (ie use of official

data self-report data observational data) (eg Curry 2000 Decker 1996 Deschenes and

Esbensen 1999 Esbensen et al 2001 Gordon et al 2004 Katz Webb Schaefer 2000 Webb

Katz Decker 2006) or research setting (ie North America Europe Asia South America

Australia) (Klein and Maxson 2006) As a consequence it should not be surprising that

policymakers academics and community leaders have focused much of their attention on

developing responses to address community gang problems

Suppression strategies have been the favored public policy response to gangs since the 1980s

(Venkatesh 1999 Spergel et al 1995) Suppression strategies typically rely on focusing

criminal justice resources on gang members through such practices as targeted and enhanced

police patrols intelligence databases vertical prosecution and enhanced sentences for those

convicted (Katz and Webb 2006) Suppression strategies are based on deterrence theory and

are founded on the principal that swift certain and severe penalties for those involved in gangs

and gang activities will necessarily result in fewer individuals joining gangs and will deter people

from engaging in gang crime (Klein 1995) By the early-to-mid 1990s as gang problems

continued to proliferate and policymakers sought alternative gang control strategies gang

prevention programming flourished (Papachristos 2013) Gang prevention programs were

aimed at the general youth population or focused on at-risk youth or neighborhoods Gang

prevention programming was based on the premise that by reducing risk factors and increasing

2

protective factors prevention specialists could inoculate youth from gang membership

(Esbensen 2000) Prevention programming has most frequently come in the form of school-

based prevention programming such as the Gang Resistance Education and Training (GREAT)

program where students are exposed to a curriculum designed to reduce both participation in

gangs and gang crime as well as the SolucionES3 program which engages youth and their

families through integrated community-based and school based programs

While a burgeoning body of literature has developed describing the assumptions issues and

effectiveness of suppression (eg Decker 2003 Decker and Reed 2002 Katz and Webb 2006

McCorkle and Miethe 2002) and prevention strategies (Esbensen and Osgood 1997) much less

attention has been given to examining gang intervention programming This might largely be

the consequence of the absence of such programming over the past 30 to 40 years By the

1980s policymakers no longer believed that social intervention approaches were an effective

strategy to control gangs and gang problems Although gang intervention strategies took many

forms they were based on the assumption that gang membership is the by-product of a socially

deprived community and that the values and norms of gang youth can be influenced and

directed toward those of mainstream society As such intervention programs often focused on

diverting youth from gangs or sought to minimize the consequences of gangs and gang activity

Gang intervention strategies include crisis intervention dispute resolution street-level

counseling and youth outreach (Spergel 1995) However many policymakers community

organizations and academics have argued that such approaches not only did not reduce gang

activity but may have lead to increased group cohesiveness which in turn may have lead to

increased delinquency (Klein 1971 Spergel 1995) More recent research has yielded similar

results For example a number of studies examining replications of Chicago CeaseFireCure

3 SolucionES is a Global Development Alliance in which $20 million is provided by USAID and SolucionES Alliance

members will contribute an additional $22 million dollars during a 66-month period starting July 2012 SolucionES is being implemented by an Alliance of five leading Salvadoran non-profit organizations who have come together to prevent crime and violence The Alliance members are Fundacioacuten Nacional para el Desarrollo (FUNDE) Fundacioacuten Salvadorentildea para la Salud y el Desarrollo Humano (FUSAL) Fundacioacuten Crisaacutelida (known locally as Glasswing) Fundacioacuten Salvadorentildea para el Desarrollo Econoacutemico y Social (FUSADES) and Fundacioacuten Empresarial para el Desarrollo Educativo (FEPADE) Together these five organizations have notable expertise in education health community development economic development research and youth leadership they are bringing their combined synergy and strengths to prevent crime and violence in El Salvador More information on social violence prevention programming is presented in Annex E including a summary of the violence prevention initiatives being implemented by SolucionES

3

Violence which relies heavily on crisis intervention dispute resolution street-level counseling

and youth outreach have found these strategies to either be ineffective or worse increase

levels of violence (Fox et al 2014)

One type of gang intervention gang truces however has received little attention in the

literature This is somewhat surprising given its use and public claims of effectiveness The

purpose of this report is to systematically examine the effectiveness of gang truces Gang truces

have largely been a ldquoblack boxrdquo that is little is known about the frequency of their use

conditions that give rise to them the role of third parties in brokering gang truces the

transformative effects of truces and the effectiveness of gang truces In this report we review

prior research on gang truces and present evidence on the effectiveness of gang truces that

have been implemented in El Salvador Jamaica and Honduras

Theoretical and Policy Rationales for Gang Truces

Much of the concern about gangs over the past twenty years has been their close association

with violence Qualitative and quantitative research has repeatedly found that gangs and gang

members are involved in high levels of serious violent offending Decker (1996) attributes the

gang-violence relationship in part to the collective and normative structure of gangs which

supports and encourages the use of violence both preemptively and in retaliation He further

explains that gang membership encourages the use of violence in retaliation against threats

and attacks whether actual or perceived which results in patterns of inter-gang conflict

characterized by escalating violence Related Cooney (1998) points out that gangs are not all

that different from ldquowarrior societiesrdquo He argues that while gangs are violent the violence

that they engage in is different than that engaged in by non-gang members in that it commonly

takes on a feud-like dimension A perceived slight violation of turf or other disrespectful

action might invoke a shooting which in turn evokes a retaliatory shooting which in turn

results in another retaliatory shooting creating a self-perpetuating cycle of violence

Early research and theoretical work examining gangs and gang conflict suggested that much

violent gang behavior was the function of status management (Thrasher 1927 Whyte 1943

Short and Strodtbeck 1965) Borrowing from the sociology and social psychology literature on

4

impression management theory gang scholars hypothesized that youth place significant

emphasis on image managementmdashseeking to impress their peers and limiting the potential to

embarrass themselves in front of others (Hughes 2005) These early gang theorists postulated

that violence is an instrument used by gangs and gang members to achieve manage and

protect status They contend that gang members often make decisions to become involved in a

violent conflict based on rational processes that weigh the immediate loss or gain of status

within the gang against the relatively small probability of being formally sanctioned by officials

within the criminal justice system (Hughes 2005) A number of researchers have examined the

relationship between status considerations and gang violence and have found the association

to be particularly robust regardless of gender (Campbell 1991 Giordano 1978) ethnicity

(Anderson 1990 1998 Vigil 1988) and location (Jankowski 1991 Spergel 1995) Hughes

(2005) notes that the centrality of status consciousness by gang youth may account for the

existence of the facilitative gang effect that gang scholars have repeatedly observed (ie Gangs

increase the amount of delinquency individuals are involved in beyond that of a group of

delinquent peers (Gordon et al 2004 Thornberry et al 2003)

Another micro-social factor associated with gang violence is group cohesiveness While our

understanding of the relationship between gang cohesiveness and violence is underdeveloped

some scholars have reported a strong relationship between the two In particular Klein and his

associates (1971 1967) and Lucore (1975) have reported that increases in gang cohesiveness

lengthens periods of gang membership and increases members participation in gang crime and

violence Cooney points out that there are strong relational ties between gang members that

necessarily result in increased cohesiveness among members For example gangs are

comprised of neighborhood youth who share common cultural and economic experiences and

have often known one another for lengthy periods of time He also points out that gangs are

groups that have strong self-proclaimed and formalized identities (eg colors symbols names

monikers) and have at least some organizational structure Their sense of group is maintained

by their common understanding of their members and friends and their attachment to their

territory (or neighborhood) Decker (1996) notes that the relatively high level of group

5

cohesiveness exhibited by gangs facilitates both collective behavior and liability (For exception

see Hughes 2013)

Accordingly both micro-social factors (ie status management gang cohesiveness) serve to

augment levels of gang violence and make it difficult for third parties to intervene Violence

within the context of gangs serves as a form of informal social control Gangs and gang

members cannot seek assistance from legitimate institutions of social control to solve conflicts

because they would risk losing status (Anderson 1999) Likewise the collective nature of the

gang not only increases potential offenders and targets of violence but also facilitates at the

group level the need for retributive justice on the behalf of injured members Moreover gang

members desire to impress others with their commitment to the group and use violence to

demonstrate their commitment to their group and to increase their status within the group All

of this results in an increased cycle of gang conflict and violence4

Over the past several decades residents and policymakers have expected the police to address

violent gang conflicts However there are several limitations to the police response to gangs

First as noted above many of the same factors that are associated with violent gang conflict

also limit the effectiveness of the police to have an impact on violent gang conflict Gang

members are not going to contact the police to resolve a conflict because it could result in loss

of status and expose them to the police discovering illegal activities that they are involved in

(Katz 2001 Katz and Webb 2006) Related citizens in neighborhoods with gang problems are

also reluctant to call the police out of fear of gang reprisals (Katz 1997 Webb and Katz 2003)

or because they have a poor perception of the police themselves (Katz Choate and Webb

2002) Third the police response to gangs in most communities is a reactive response to a

specific incident after it has occurred rather than a proactive response to intervene in ongoing

disputes between gangs (Katz and Webb 2003 2006) Most police agencies simply do not have

4 There is anecdotal evidence for these hypotheses Hughes and Short (2005) through field observations and interviews with members of 20 Chicago gangs examined the micro-social context of violent-related incidents Specifically they were interested in the specific social conditions that precipitated violence Their analysis revealed that gang disputes associated with retaliation escalated into violence significantly more often (ie 55 of the time) than disputes associated with normative order violations or an identity attacks which only resulted in violence roughly 33 percent of time

6

the intelligence networks required to intervene in gang conflicts until after they have risen to

relatively high levels (Katz 2003) Fourth police suppression strategies have been linked to

increasing gang cohesiveness and possible increases in gang crime (Klein 1995)

Some policymakers and community activists have proffered that an alternative to reliance on

formal mechanisms of social control such as the police is the gang truce (Spergel 1995) The

goal of a gang truce is to reduce or even eliminate violent conflict between those gangs that are

warring with one another As such compared with other strategies that often seek to reduce

general levels of crime a gang truce has the very specific goal of reducing violence between

two or more gangs that are in conflict with one another Unfortunately the literature provides

little guidance on the theoretical assumptions of why a gang truce should reduce inter-gang

violence Henderson and Leng (1999) hypothesize that at the root of gang truces is the notion

that they involve the renegotiation of existing norms within and between gangs The authors

claim that as a violent dispute escalates between gangs leaders and members are placed in the

situation of appearing weak to both members of their own gang and to members of the rival

gang if they do not respond with the appropriate amount of force or if they were to suggest a

peaceful resolution to the dispute Accordingly Henderson and Leng argue that as the cycle of

violence escalates between two or more gangs behavioral norms shift toward the increased

valuation of violence to resolve the conflict because it is the only option readily available to

them A gang truce on the other hand which is often mediated by a third-party is believed to

break the cycle of violence by providing the gangs involved in the dispute with a cooling-off

period (Spergel 1995) In the interim period new norms of expected behavior within and

between gangs are established In other words the cooling-off period is believed to recalibrate

norms of behavior that are more consistent with the security interests of the gang and its

members (Henderson and Leng 1999)

For many of the reasons stated above a number of communities have participated in gang

truces Gang truces have been observed in the United States Central America and the

Caribbean and gang leaders in other nations are considering its implantation (Fahah 2012)

Unfortunately little is known about gang truces We do not know when they come into

consideration how they are implemented whether they decrease increase or even have a

7

significant impact on violence In the below section we systematically review the existing body

of literature on gang truces Appendix A describes the methodology we employed to conduct

the systematic review

Prior Research on Gang Truces

Little research to date has examined the effectiveness of gang truces and of the research that

has been conducted has been restricted to the field of public health The most celebrated

truce to date was Philadelphiarsquos House of UMOJA 1974 gang truce In preparation of the truce

gang leaders in prison were called upon for their support along with key community leaders

from churches businesses schools and the police When UMOJA called for the gang summit it

was reported that 500 gang members and 75 of Philadelphia gangs attended By the

conclusion of the gang summit a 60-day truce was announced which reportedly resulted in no

gang members being killed over the 60-day period However no systematic evaluation of the

gang truce was conducted (Woodson 1981)

The first attempt to evaluate a gang truce to our knowledge was conducted by Cotton (1992)

who examined the results of a gang truce in South Central Los Angeles between the Crips and

Bloods Data provided by the police department indicated that over the 6-week period when

the truce took place drive-by shootings decreased by 48 compared to the same 6-week

period the prior year decreasing from 162 to 85 Likewise gang-related homicides dropped by

62 from 26 to 105

Ordog et al (1993 1995) examined the effects of a gang truce in Los Angeles using emergency

room admissions data Specifically the authors examined changes in the daily and monthly

number of gun shoot wound (GSW) emergency room admissions before during and after a

gang truce The catchment area for the emergency room was 100 square miles The authors

noted that while they were able to clearly identify the date the gang truce began because of

the media coverage that it received there was no specified date that the truce ended and as a

5 The authors did not report whether the drop in gang-related homicides was citywide or in the South Central Los Angeles neighborhood where the gang truce took place

8

consequence after 12 weeks the authors regarded it as called off for evaluation purposes

because it was no longer being discussed in the media and gun shot wounds began to increase

Student t-tests were used to examine changes before during and after the gang truce Ordog

et al (1995) reported that there were approximately 7 GSW admissions per day the 12-months

preceding the truce compared to 45 GSW admissions per day during the gang truce and 126

GSW admissions per day in the 11-months following the gang truce The authors concluded

that their analysis ldquoclearly showed that the institution of a gang truce had reduced the number

of GSW victims seen in an hellip inner city Level I trauma centerrdquo (p 419)6

However it is important to point out that while the gang truce in Los Angeles did decrease

homicides by about 35 percent for the first three months it then doubled in months four

through eleven compared with the pre-truce period - a pattern that the authors did not

discuss Similar findings were reported in Trinidad and Tobago (Maguire Katz and Wilson

2013) where it was determined that homicides declined for a brief period of time (again for

about three months) but then increased substantially over the long term (12 months) These

results suggest that gang truces may produce short-term benefits yet result in long-term

adverse consequences

While the research examining gang truces shows their potential promise and their potential for

greater harm we believe that the findings should be viewed with caution for two reasons

First some of the implicit theoretical assumptions that gang truces are built on may not be

accurate Prior research on gangs suggests that they have limited organizational structure

(Decker Katz and Webb 2008 Decker Bynum Weisel 1998) and have few formal

mechanisms to influence member behavior For example most gangs do not have formal

leaders do not require members to pay dues and members do not make contributions to the

gang for the purpose of developing the gangs infrastructure (ie guns housing etc) A number

of academics have pointed out that ldquosophisticated gang organizations is still largely a product of

the self- or organizational-interested musings of gang leaders certain police officials academic

6 Similar results were presented by Ordog et al 1993 using the same data but analyzing it over a shorter period of time and aggregating the data by month instead of by day

9

researchers media reporters based on very limited hard datardquo (Spergel 1995 79-80)

Therefore even if a truce was successfully negotiated between members of gangs in conflict

much of the academic research suggests that gang leaders do not have enough control over

members to enforce a truce Related prior research on gangs and gang members suggests that

they are primarily comprised of young people with few ldquodiplomaticrdquo skills (Henderson and Leng

1999) The very same organizational and normative features of gangs that result in gang

violence (ie often bastions of young minority male street youth) are those same features that

most likely limit the likelihood of a gang having the capacity to abide by a gang truce In sum

implicit in the theoretical assumptions of a gang truce is that gangs have the organizational and

cultural capacity to create and maintain (at least for a short while) a truce which is inconsistent

with much prior research on the organizational characteristics of gangs

The second reason we believe that prior evaluations of gang truces should be viewed with

skepticism is that they have relied on relatively weak evaluation designs For example prior

studies examining gang truces have not incorporated the use of comparison areas or control

groups The causes of reductions in gang violence found in previous evaluations might be

many For example the Los Angeles riots took place just before the gang truce that Ordog et al

evaluated Zinzun (1997) reported that gang culture and violence changed briefly but abruptly

following the riots because gangs and gang members in part redirected their anger and focus

toward the police As such the decline in GSW admissions may have been the consequence of

an overall city-wide decline in gang violence in the wake of the LA riots Related prior research

has relied on fairly broadly defined outcome measures such as GSW admissions and general

levels of drive-by shootings Such outcome measures lack specificity in terms of attributing

gang violence to the specific gangs involved in the gang truce Evaluation designs used in

previous studies have also lacked specificity in terms of the catchment area where violent

activity was labeled as gang related For example Ordog et alrsquos (1993 1995) outcome measure

included all emergency room admissions for GSW However the emergency room received

patients from a 100 mile square area surrounding the hospital an area that was most likely

much larger than the gangsrsquo territories involved in the truce

10

Additionally prior research examining gang truces has not examined the processes involved in

the creation of the gang truce No context has been provided in terms of the factors that lead

to the gang truces whether the gangs were pushed or pulled into truces whether outside

parties helped mediate the truces or whether on-going mediation was required to maintain the

truce In other words we still do not know the processes related to the formation of a

successful gang truce Instead prior research and evaluation has treated the gang truce much

like a black box where it is described in very general terms but its details are not revealed

Likewise prior evaluations have not examined the effects of a gang truce beyond its impact on

gun violence A number of prior researchers have argued that mediating such activities serves

to legitimize gang leaders increases gang identity for members and results in greater group

cohesion (Haskell and Yablonsky 1982 Klein 1995) As such we do not know if there are any

unintended consequences resulting from gang truces due to the methodologies chosen for

prior research and evaluation Finally prior research on gang truces has not relied on more

sophisticated statistical analysis that has the capacity to not only discern overall trends but also

cyclic patterns outliers and turning points

THE PRESENT STUDY

The present study seeks to understand the effectiveness of the gang truce when negotiations

occur with or between gangs and with government institutions or civil society organizations

We have briefly reviewed the theoretical and policy rationales for gang truces and the

literature and prior research on this form of gang intervention Next we present three case

studies one from El Salvador another from Jamaica and another from Honduras Each offers

evidence about their impact on violence and lessons learned that are unique to their particular

circumstances In conclusion the case studies are considered together and we present their

collective policy implications

11

Case Study A Gang Trucemdash The Salvadorian Experience

Over the past two years the Salvadoran gang truce brokered by an array of local policymakers

and international development organizations has attracted national and international

attention Violence in El Salvador has been at an all time high with a rate of 692 homicides per

100000 population in 2011 (UNODC 2014) making El Salvador one of the most violent nations

in the world (Parkinson 2014) Since 1992 government and non-government actors have been

responding to the rising tide of gang violence using traditional suppression strategies such as

increased policing legislative changes and more severe prison sentences These traditional

mechanisms of formal social control however were proving ineffective if not counter-

productive (Perez 2003)

In response to the inadequacy of traditional strategies stakeholders altered their course in an

effort to radically reduce gang violence in the nation Members of the Funes administration led

a group of negotiators comprised of the Catholic Church a former congressman and the

Organization of the American States (OAS) to help frame the conditions for a possible truce

between the MS13 and 18th Street gangs (Umana de Leon and Tager 2014) In March 2012 a

truce was reached The goal of the gang truce was to reduce violence specifically gang-involved

homicides Included in the terms in exchange for the gangs acting to reduce homicides certain

incarcerated gang members were to be transferred to lower security prisons to receive

sentence reductions and special visitation privileges and to be permitted more communication

with those outside of the prison for the purpose of conducting crisis interventions to mitigate

the violence (Salanegra 2012)7 The gang leaders also agreed to no longer recruit children into

their gangs reduce violence against women give up a small number of guns and continue to

participate in negotiations (Seelke 2014 11-12)

7 It is important to note that there were other discussions that took place during the mediations that included as a local publication noted ldquoProhibition Act Gang send the army to barracks end the police operations in controlled areas by gangs repeal of the figure of the witness criteriado (with criteria of opportunity or ldquowitness under a plea agreementrdquo) and a number of improvements in the quality of life of the inmates and its familiesrdquo Source httpwwwsalanegraelfaronetes201209cronicas9612

12

The present case study seeks to understand the role of negotiations with or between gangs and

government institutions or civil society organizations and their impact on violence in El

Salvador The specific objectives of this case study were to (1) Identify and document the

negotiation processes with or between gangs This includes but will not be limited to

identifying the actors involved in the negotiations the goal(s) of the negotiations and

strategies employed to carry out negotiations (2) Identify the impact of negotiations with

andor between gangs on violence and other forms of criminality and (3) Present conclusions

about the strengths and weaknesses of negotiations and their potential for positively or

negatively reducing crime in Salvadoran society

The Salvadorian Gang Truce

While the origins of the Salvadorian gang truce are somewhat unclear some accounts suggest

that in the fall 2011 the Minister of Security (then David Munguia Payes) mentioned the idea of

starting a dialogue between the gangs to Raul Mijango (a former congressman) Shortly

thereafter Mijango spoke to Monseigneur Fabio Colindres the military chaplin about the idea

They started to communicate with gang leaders in January 2012 From the beginning the

Minister of Security stated that he would not personally communicate with gang leaders and

noted that the official position of the government was not to negotiate with offenders But he

did allude to the fact that the gang truce was part of a new strategy to address the nationrsquos

gang problem (Archibold 2012) The Salvadorian gang truce was multi-dimensional involving a

varying number of actors communication styles and tactics In the section below we discuss

the parties involved in the truce the negotiation process and the strategies used by the gangs

and government in furtherance of the truce

Parties involved in the truce There were a series of persons groups and other entities who

directly or indirectly participated in the negotiation process whether they were negotiators or

collaborators From the beginning negotiations between the MS13 and 18th Street gangs were

considered confidential and were held in strict confidence between the gangs and the

negotiators Beginning in early 2012 the first two mediators to serve were a former

congressman (Rauacutel Mijango) and a priest (Fabio Colindres) These individuals were able to gain

13

the trust of the gangs and also had open communications with the government and

community

As the process moved forward however church leadership changed its initial position and

withdrew its support of the truce This left the priest to participate as a private individual no

longer serving as a representative of the church The mediators had some early success

however as the mediators changed the lines of communication became less clear and less

reliable Still they played an important role in the process Finding individuals whom the gangs

would trust and who could speak for government and community institutions was challenging

Further as government ministers and presidents changed it became less clear whether or not

the government was participating in the dialogue between various parties much of the

confusion in negotiations was related to understanding the governmentrsquos official role in the

process

In the beginning the negotiating parties included the two mediators and MS13 and 18th Street

the two predominant gangs in El Salvador MS13 is the largest gang in the country with about

250 cliques throughout the nation For the most part its first- and second-generation leaders

were in prison There are municipalities in which only MS13 operates and therefore its

members never clashed with 18th Street adversaries MS13 members are largely involved in

extortion violence and intimidation of the public The MS13 leadership has been shown to

have greater control of its organization than most its members have been more disciplined

and despite its size the gang seems to have sufficient and effective communication

mechanisms between cliques and its leadership (ldquoranflardquo or ldquoruedardquo)

The 18 Street gang is somewhat smaller than MS13 Long before the truce process began

factions existed within 18th street which had become deeply involved in internal conflicts over

leadership structure Those differences had become more pronounced over time leading to the

establishment of two movements known as the ldquoSurentildeosrdquo (Southerners) and the

ldquoRevolucionariosrdquo (Revolutionaries) that participated in the peace process but under different

14

representation Despite their differences however for the negotiation process the two

fractions of 18th street established similar demands between themselves and with MS138

As noted above the governmentrsquos involvement in the truce process was never clearly

articulated and it varied with ministerial changes The Funes Administration specifically the

Ministry of Justice and Public Security (MJPS) had been engaged in the negotiation process

early on In 2013 however with the change in MJPS leadership this also changed Tensions

grew over the role of the government in the truce negotiations According to the mediators the

new Minister of Security wanted to dismantle the process and for this reason he blocked the

mediatorsrsquo entry to the prisons where they had been meeting with gang leaders The change in

government leadership brought changes in strategy As with all negotiations mutual trust was

the cornerstone When the personnel and their agendas changed so did the trust in the peace

process

Late in 2012 as negotiations began to focus on the relationship between the gangs and the

community other organizations were called upon to help facilitate this interaction A technical

committee was created to oversee progress towards the violence reduction goal this was

announced at the end of August 2012 The committee moved forward with support from the

Organization of American States (OAS) It coordinated actions intended to improve the living

conditions of incarcerated persons to reintegrate into society those who had served their time

in prison and to prevent violence and provide assistance to victims of violence The committee

operated as a mechanism for coordinating the negotiation process with technical support from

the United Nations Development Program (UNDP)‒El Salvador Interviews with experts

indicated that Fundacioacuten Humanitaria with support from several organizations like Interpeace

and others had also supported the operations of the mediators and served as an important

point of reference for the gangs

In each territory where agreements were made government representatives from some

municipalities were invited to participate in the process as key implementation stakeholders

Between December 2012 and January 2013 the municipalities of Ilopango Santa Tecla

8 httpsalanegraelfaronetes201110cronicas5645

15

Sonsonate and Quezaltepeque became among the first to become part of what later became

known as the Sacred Municipalities (aka violence free municipalities) By November 2013

several other municipalities had become part of this group La Libertad Apopa San Vicente

Puerto El Triunfo Nueva Concepcioacuten Ciudad Delgado and Zacatecoluca

The gangs and the government offered several incentives to one another The gangs offered a

number of actions to reduce the incidence of crime (ie cease of hostilities zero homicides

stop extortions robberies and recruitment of children and youths into gangs peace at the

schools) and the government offered to guide public investments in social policies prioritizing

participating municipalities

The truce process was the product of a convergence of vested interests from different sectors

Several earlier attempts at negotiations had failed likely because of a lack of perceived

authenticity on the part of the actors All parties involved in this truce process came to the

table with well-defined group interests and concerns All sides were experiencing fatigue from

the long history of violence Years of fighting deaths and violence had led the mostly

incarcerated first generation those who were still living and who continued to serve as gang

leaders to seek new ways of exercising social power and influence in prison and the

community

The negotiation process In El Salvador the negotiation process between gangs formally began

on March 8 2012 by a decision of the Government to favor the transfer of gang leaders

imprisoned out of the countryrsquos only maximum security prison located in the municipality of

Zacatecoluca That decision was made by the Minister of Justice and Public Security as part of

an agreement between a team of mediators who served as advisors to the Ministry and gang

leaders primarily from MS13 and 18th Street The negotiations were referred to as a ldquotruce

between gangsrdquo and were characterized by the mediators as ldquoa peace processrdquo or a ldquocease of

hostilities between gangsrdquo

Publicly the gang truce was known as an agreement between gangs and not between gangs

and the government because of fear of how those outside the negotiations might interpret the

governmentrsquos actions From the beginning therefore the government never fully

16

acknowledged its participation in the peace process (through representatives) with the gangs

and that reticence was a major impediment throughout the process This was largely because

of the difficulty in communication between the multiple parties which was done through

mediators and necessarily resulted in additional logistical complexity throughout the

negotiations Regardless it is important to recognize that the government played a decisive

and central role in the origin facilitation and promotion of the peace process

In terms of processes interviews with gang leaders indicated that they would communicate

their ldquodemandsrdquo to the mediators who in turn would communicate the gang leaderrsquos demands

to the government The same process was used for the government to communicate with the

gang leaders After an agreement was reached the parties would make proclamations that

emphasized the agreed upon terms of the negotiations These often took the form of public

statements made through local newspapers For example there were more than twenty public

statements made by the gangs where they clearly articulated that their intention was to

reduce the harm that their acts of violence were causing themselves and the communities (For

a detailed discussion of the public statements see Appendix B) In these public statements the

gangs acknowledged the need for their groups to change Both the mediators and the gangs

made clear that the gangs expected the government to respect and respond to their demands

given the good-faith actions that the gangs had already taken

The government on the other hand did not make any formal public statements about their

end of the agreements reached through the negotiations From the outside it appeared that the

government did not develop a formal or consistent policy regarding the truce and instead

assumed that they could achieve a reduction in homicides through the negotiation of increased

prison privileges for gang leaders At the close of the Funes administration the peace process

and negotiations initiated within its framework had reached a plateau There they remained

waiting for the new government to step in and take action as President-elect Salvador Saacutenchez

had announced his support for the truce during the electoral process Sanchez however

withdrew that support when he assumed the presidency in June 2014 Indeed under the

recently elected Sanchez administration the government changed course and became less

17

willing to engage the gangs believing that the government should not negotiate with criminal

organizations

Indeed it should be noted that responses resulting in increased gang influence have been a

concern over the past several decades--from governmental neglect of the problem in the

1990s through the period of the mano dura and super mano dura of the early 1990s to

2000 and to the present gang truce Many have asserted this claim as truth (ie some

responses will increase the influence of gangs) especially in public discourse This study

however does not directly address this point which may or may not be valid9

Strategies employed during negotiations between the gangs and the government Over the

course of the negotiations between the gangs and the gangs and the government at least two

strategies were employed At the beginning of the negotiations strategies were implemented

similar to that of the historic Salvadorian peace accords Each of the parties with the support of

mediators reached agreements and achieved a resolution to their conflict This model

although useful overlooked an important difference between the current violence in El

Salvador and the civil war of the 1980s In a conflict of civilians with the State the legitimacy of

the ldquoadversariesrdquo arises by the need of the state to recover the rule of law It is certainly

possible to reach peace agreements that resolve conflict between gangs but in the current

case with regard to the government there was no legal or policy justification for executing the

truce process On the contrarymdashgovernment negotiations with a criminal group are relatively

rare in modern democratic societies The lack of a legal or policy framework to work from

limited the governmentrsquos ability to be transparent in its response to gangs and may have

undermined its legitimacy with the public

The second strategy employed by the parties was based on reciprocation and cooperation

between actors Early on it was determined by both sides that if one of the parties abandoned

the peace process or did not hold up its end of an agreement the other party would no longer

participate in the negotiation process (Axelrod 1986) In retrospect during the course of the

peace process the gangs implemented agreed-upon terms and positioned themselves as valid

9 We would like to thank one anonymous reviewer for bringing this issue to our attention

18

partners with the government able to negotiate for what is needed by the government in

exchange for what the gang needed10 For example in public statements the gangs insisted

that they had made a goodwill gesture when they declared a unilateral truce and stopped their

involvement in violence The government responded by transferring incarcerated gang leaders

from a maximum-security prison to ordinary criminal prisons11 Next the gangs offered to

disarm the government responded with increased flexibility on control measures at those

prisons This process continued until the demands from both sides grew in a direction that

challenged each sidersquos capacity and willingness to follow through

For instance mediators through the government asked the gangs to end the extortion of

businesses and individuals which are the primary means of subsistence for Salvadorian gangs

The gangs asked that the government reciprocate by eliminating the ldquowitness under a plea

agreementrdquo from the criminal procedural law which is one of the main weapon laws used for

sentencing gang members for complex crimes such as extortion Each of these requests were

more than the parties could carry out Exhibit 1 outlines the different ldquooffersrdquo and ldquodemandsrdquo

made by each party The offers and demands are divided into those that were believed to be

relatively simple and were ldquoofferedrdquo by the party on its own volition and those that were

believed to be more complex and were demanded by the other party

Exhibit 1 Simple demands and concessions made by gangs and the government

Gang concessions Government concessions

Reduce the homicides by more than 50 Transfer the leaders to prisons with lesser security

Hand over 500 firearms to the authorities Allow visits of the children of the gang members

Reduce violent actions at schools Allow night intimate visits

Stop killings at the prisons Allow entry of fast foods

Do not murder custodians police soldiers or their family members

Facilitate the entry into the prison of gang members let out of prison

Give opportunities to some gang members to withdraw

Suspend the searches by the armed forces on persons at the prisons

10 Source httpwwwsalanegraelfaronetes201209cronicas9612 11 This was useful to facilitate the coordination and the communication between the gang members in prisons and in the streets

19

Make it more flexible for new social plans by government to be established in the communities controlled by gangs

Improve the conditions and health assistance at the prisons (eg place tvrsquos to improve mental health of inmates)

Do not burn buses do not kill bus drivers or collecting agents in buses

Allow the Red Cross and journalists to enter into the prisons

The second agenda presented in Exhibit 2 shows the more complex demands that were

not offered by one side or the other but rather were demands placed on the other party These

demands were such that they required a higher level of authority to negotiate in order to

implement the demanded action These demands were considered critical for the peace

process to continue and the delayed responses on both sides stalled the progress of the truce

and led to the parties questioning the legitimacy of the other side all these demands have yet

to be attained The mediators recognized at the end of 2012 that some of the slow progress

was related to the difficulty of making the transition from offers that could be executed

relatively simply to the more complex demands made by each party One of the gang leaders

summarized the issue ldquoWe are not looking for television sets while all of our people continue

living like shit we are not going to try to do everything that is in our power to decrease as much

violence as possible for one television set There are things we are very clear about this [points

to a television set in the room] is a right that the law grantsrdquo

Exhibit 2 Complex demands and concessions made by gangs and the government

Demands by Gangs Demands by Government

Eliminate the figure of witness under a plea agreement from the Criminal Law

Stop homicides and extortions indefinitely

Create international commission to investigate the human rights violation cases of the gang members and their family members on the part of the State (PNC and FA)

Share the information regarding the whereabouts of the brunt of extortion money whether in country or abroad

Institutionalize external and professional surveillance regarding the behavior of the PNC as regards investigations and gang member arrests

Progressive dismantling nation-wide of the clique structures and turfs

Army should stay in its barracks and definitely suspend their participation in public security tasks

Permanently suspend the orders to murder State security and justice agents and their family members (police soldiers judges prosecutors)

20

Clearing of judges prosecutors and police involved in corruption cases against youths in conflict with the law

Permanently suspend the murders extortions and harm caused to public transportation resources and their workers

Maintained a sustained improvement on the conditions of the prisons

Permanently suspend the recruitment of children and youths and hostilities to educational centers

Guarantee working opportunities for the gang members and their family members through specific programs at the municipalities

Share the information about providers of drugs and illegal arms

Television sets were one of the concessions made by the government from the simple list The

gang leader recognized that their intentions in the negotiations extended beyond improving

prison conditions however the government did not have support nor established mechanisms

for carrying out tasks derived from the more demanding list Further as the government

transitioned to new presidential leadership many of the more simple concessions had already

been made and the new government would have to address more complex demands if

negotiations were to continue

The Salvadorian gang truce is remarkable for several reasons First a number of policymakers

and researchers have claimed that the truce saved a large number of lives and was perhaps the

most successful gang truce in the Western Hemisphere Second the Salvadorian gang truce is

somewhat unique in that it might have lasted substantially longer than any other successfully

negotiated gang truce Understanding the temporal impact of the truce is important to

understanding its future potential Third the Salvadorian gang truce is important because a

number of other counties have sought to replicate it For example following the perceived

(and perhaps real) success of the Salvadorian gang truce the nations of Honduras Belize and

Guatemala instituted or tried to institute a similar type of truce Understanding the impact of

the Salvadorian gang truce will further help us understand the potential impact of such a

process on violence in these other nations

Methods

We examined the impact of the gang truce by merging four separate data sets First data from

the 2007 El Salvador Population and Housing Census provided municipal level measures of

21

percent unemployed percent male aged 10-29 percent female headed households ethnic

composition in-migration income percent urban percent households rented and percent

who had a high school education These data were obtained directly from El Salvadors Ministry

of Economy Second municipal level crime data (ie homicide extortion rape theft robbery

and auto theftrobbery) were provided by the National Civil Police (PNC) by month and year for

the period between January 1 2010 and June 30 2014 Disappearance data was also provided

by the police aggregated by year and municipality Third police gang intelligence data was

provided by the National Civil Police (PNC) Specifically we received the number of police

recorded MS 13 and 18th Street gang members by municipality in 2011 the latest year for

which these data were made available Last we acquired 2011 prison gang intelligence data on

the number of imprisoned MS 13 and 18th Street members by municipality from the

Salvadorian National Bureau of Prisons All four data sources were linked together using a

unique municipality identification number and month Collectively they provide an

opportunity to examine the impact of the gang truce in El Salvador and understand several

competing explanations for any changes that might have occurred

The geographic unit of analysis for the present study is the municipality According to the 2007

El Salvador Population and Housing Census there were 14 departments divided into 262

municipalities (the equivalent of a county in the USA) in El Salvador Of the 262 municipalities

ten (38) of the municipalities were eliminated from the analysis because of missing data

Measures

Dependent variables The dependent variable in the present study is the monthly number of

homicides in each El Salvadoran municipality expressed as the number of homicides in

municipality i at time t Homicide data were collected from January 1 2010 through June

30 2014 for a total of 54 months This includes data for a period of 26 months prior to the gang

truce and 28 months following the implementation of the gang truce As shown in Exhibit 3

there were a total of 14148 homicides over the study period with each municipality averaging

371 homicides per month

yit

22

Exhibit 3 Descriptive statistics

n mean sd

Urban Opportunity Factor 252 111 95

male 10-29 years old 252 1960 171

female-headed household 252 3433 499

unemployed 252 1177 686

Racialethnic heterogeneity 252 17 13

of prisoners MS13 252 1579 3960

of prisoners 18th Street 252 1268 4096

of prisoners MS13_spatial weight 252 1579 3960

of prisoners 18th Street _spatial weight 252 1268 4096

MS13 on the street 252 4146 10646

18th Street on the street 252 2513 8622

Violence free municipality 252 04 20

Truce 14148 371 783

Monthly homicide rate per 1000000 14148 51 49

Independent variables Several independent variables were included in our model Our

measure for the intervention is a dummy variable in which the value of 0 is used to represent

the pre-intervention months and 1 represents the post-intervention months We also

included a dummy variable for each community that participated in the violence-free

municipality initiative (sanctuary municipalities) to address issues of external validity (eg

multiple treatment inference) In other words municipalities that participated in the

violence-free initiative might have experienced or participated in something that either

enhanced or detracted from the impact of the larger gang truce The initiative called by some

the second phase of the gang truce took place about 8 to 9 months after the original gang

truce was negotiated in March 2012 In participating sanctuary municipalities gang members

agreed to stop violence and crime in exchange for a reduction in police operations and night

raids12 The municipalities that participated in the initiative included Santa Tecla

Quezaltepeque La Libertad Ilopango Ciudad Delgado Apopa Sonsonate San Vicente

Zacateatecoluca Puerto El Triunfo and Nueva Concepcioacuten

12 See httpwwwcispesorgblogviolence-free-cities-inaugurated-as-second-phase-of-gang-truce

Tt

Si

23

We included several independent variables related to the presence of gangs by gang for each

municipality Specifically we included count variables on the number of MS13 and 18th Street

gang members who were on the streets and in prison by municipality As noted before the

gang truce was made between a relatively small number of imprisoned gang leaders from both

MS13 and both fractions of 18th Street who agreed to stop street-level gang violence on the

condition that they would be transferred to lower security prisons and granted special

privileges

We wanted to understand the impact of the truce in the context of gang presence

Municipalities with high numbers of MS13 and 18th Street members whether they were in

prison or on the streets should have experienced a greater reduction in homicides because of a

greater span of control over these communities Accordingly municipalities with low numbers

of MS13 and 18th Street members should have experienced a lesser reduction in homicides

because of a limited span of control over these communities As Exhibit 3 (above) shows on

average municipalities had about sixteen MS13 and thirteen 18th Street members in prison

and forty-one MS13 and twenty-five 18th Street members on the street We examined whether

gang members in El Salvador were randomly distributed and found that there was spatial

clustering in the number of MS13 and 18th Street members who were in prison In order to

control for the clustering we included the spatial lag (weight) of the MS13 and 18th Street

members in prison

Control variables A series of control variables known through prior research to be related to

violence in communities were added using the census data We began with municipal-level

measures of percent unemployed percent male aged 10-29 percent female-headed

households ethnic composition in-migration income percent urban percent households

rented and percent with a high school education Initial analysis found that five community

variables were highly correlated and loaded on the same factor As seen in Exhibit 4 these

included the percentage of persons who had moved there from another municipality (in-

migration) average income in the municipality percent of the population living in an urban

area percent of houses that are rented and percent of residents with at least a high school

education We labeled this factor Urban Opportunity

24

Exhibit 4 Factor loadings for urban opportunity

Variables Factor

loadings

In Migration Percent of population moved in from another municipality 0638

Income Average monthly income per household (colones) 0886

Urban Percent of population that is urban 0845

Rented Percent of households rented 0761

Education Percent of residents who have at least a high school education 0742

Percent of variance 6071

Eigenvalue 341

Extraction method Principal Axis Factoring

We also calculated a measure of ethnic-heterogeneity from relevant census data Ethnic

heterogeneity which varies from 0 to 1 was calculated by taking one minus the squared

proportions of the population in each ethnic group (White Mestizo other) As with the

percentages of male population that is 10-29 years old female-headed households and

unemployed persons ethnic heterogeneity has consistently been associated with violence in

general (Kubrin and Weitzer 2003) and with gang violence in particular (Katz Maguire and

Roncek 2002 Rosenfeld Bray and Egley 1999)

Analytic Strategy

National-level analysis Our analysis began with a simple t-test of the mean number of

homicides at the national level before and after the truce providing the most basic omnibus

test of an effect We then presented our time series model to estimate the effect net of

seasonality and temporal trends This analysis included a set of simple ARMA (autoregressive

and moving average) models with two lag periods and one period of a moving average Initial

analysis of the number of homicides by month indicated that partial temporal autocorrelations

existed for two lags The first model used only data prior to the truce This model included a

linear time trend variable and dichotomous variables for each month (except January) We used

this model to forecast the expected number of homicides for the truce period This series of

analyses was for illustrative purposes only as the number of data points used in the forecast

25

was too small to provide meaningful confidence intervals The second ARMA model employed

all data from January 2010 through June 2014 and included a dichotomous indicator for the

truce period The effect of this variable was the average change in the number of homicides

net of seasonality (months) and temporal trends

Municipality multilevel models The analysis at the municipal level presented four challenges to

a typical regression model First since our outcome has a highly skewed distribution with

varying levels of exposure (population) we employed a generalized linear model to capture the

correct distribution In this case we employed a negative binomial distribution rather than a

Poisson model due to the over-dispersion in our outcome created by analyzing monthly

municipal datamdashthat is there were several months and municipalities where no or very few

homicides occurred

The second challenge was our need to measure the between-municipality variation of pre-truce

homicide rates and the program effect A fixed-effects model would have been inappropriate

because it would not have allowed us to estimate these variance components Thus we

employed a multilevel random effects generalized linear model

The third challenge was temporal autocorrelation Analysis at the national level indicated

partial autocorrelations in the first and second lags Although a generalized model does not

allow for auto-correlated residuals we addressed this by entering in the model two lags of the

homicide rate

The fourth and final challenge to this estimation was spatial autocorrelation An examination of

the Moranrsquos I and autocorrelation coefficients (exhibit not shown) indicated a low level of

autocorrelation for each month but many months were still statistically significant Thus we

estimated spatial lags of the homicide rate and entered them into the model

Accordingly we analyzed the data using random effects negative binomial models regressing

the rate of homicide on various months on predictors with both temporal and spatial lags For

clarity we present the model using HLM (Bryk et al 1996) notation At the first level (time) we

estimated the log number of homicides using a negative binomial distribution with predictors

26

that included the truce period calendar month linear month trend a

one period lag of the homicide rate a two-period lag of the homicide rate a

spatial lag of the homicide rate and an over-dispersion parameter which has a

Gamma distribution of Thus the final level 1 model is

At the municipality level (level 2) the intercept is a function of the presence of the

Violence Free Municipality program the log number of MS prisoners the log

number of 18th Street prisoners the log of the spatial lag number of MS prisoners

the log of the spatial lag of the number of 18th Street prisoners the log of the

number of MS members the log of the number of 18th Street prisoners the

control variables detailed above and the random effect Thus the level 2 model for the

intercept is

Also at the municipality level (level 2) the effect of truce is a function of SAFE the

log number of MS prisoners the log number of 18th Street prisoners the log of

the spatial lag number of MS prisoners the log of the spatial lag of the number of 18th

Street prisoners the log of the number of MS members the log of the number

of 18th Street prisoners and the random effect Thus the level 2 model for the

truce effect is

TRUCEt mt TIMEt

HRt-1i HRt-2i

HRti vti

Gamma1

aa

aelig

egraveccedilouml

oslashdivide

mti = expb0 i + b1iTRUCEt + lmmt

m=1

11

aring + l12TIMEt

+l13HRt-1i + l14HRt-2i + l15HRti + vti

eacute

euml

ecircecircecirc

ugrave

ucirc

uacuteuacuteuacute

exp vti[ ] ~ Gamma1

aa

aelig

egraveccedilouml

oslashdivide

b0 i

SAFEi PMSi

P18thi

PMSi P18thi

MMSi M18thi

Ci u0 i

b0 i = g 00 +g 01SAFEi +g 02 ln PMSi +1( ) +g 03 ln P18thi +1( ) +g 04 ln PMSi +1( )

+g 05 ln P18thi +1( ) +g 06 ln MMSi +1( ) +g 07 ln M18thi +1( ) + pC

C

aring Ci + u0i

b1i SAFEi

PMSi P18thi

PMSi

P18thi MMSi

M18thi u1i

27

The key parameters of our analysis are in this expression The truce effect is dependent on

the log number of MS prisoners among other characteristics Thus the percent

reduction in homicides can be calculated from this model using the following formula

which we plot below (see Quantitative Findings below)

To answer questions about the variation in the effects of the gang negotiations we predicted

best linear unbiased predictions (BLUP) of the Truce random effect which estimates the

between-community variance of the immediate effect of the truce Since we assumed that the

interventions were not evenly spread throughout communities in this study the mixed model is

appropriate to address this issue We conducted this analysis for homicide using a Stata

generalized linear mixed model using full maximum likelihood and an identity covariance matrix

of random effects

Findings

Exhibits 5 and 6 present our findings related to the number of police recorded homicides in El

Salvador from January 2010 through June 2014 The trends in the data and the results of the t-

test show that in the 26 months prior to the gang truce there were on average about 354

homicides per month compared with about 218 homicides per month in the 28 months

following the gang truce Exhibit 7 shows that the gang truce resulted in 5501 fewer homicides

than otherwise would have occurred

b1i = g 10 +g 11SAFEi +g 12 ln PMSi +1( ) +g 13 ln P18thi +1( ) +g 14 ln PMSi +1( )

+g 15 ln P18thi +1( ) +g 16 ln MMSi +1( ) +g 17 ln M18thi +1( ) + u1i

b1i

ln PMSi +1( )

1- exp g 10 + ln PMSi +1( )g 12( )eacute

eumlugraveucirc( )acute100

u1i

28

Exhibit 6 Number of homicides in El Salvador by month with ARIMA Model Fit

Exhibit 5 Change in homicides from pre-truce to post-truce

Obs Mean SD 95 confidence interval

Pre-truce period 26 35442 4200 33745 37138

Post-truce period 28 21835 6361 19369 24302

Change -13606 10638 16574

=p lt 05

29

Exhibit 7 Forecast of homicides without gang truce

Next as seen in Exhibit 8 we examined the spatial distribution of the change in the homicide

rate after the implementation of the gang truce The analysis showed that of the 252 analyzed

municipalities 243 (93) experienced a decrease in homicides however within these

municipalities there were wide variations in the degree of the decline For example of the 243

municipalities that experienced a decrease the decrease in the homicide rate varied from

about 59 percent of these municipalities experiencing a 1-74 percent decrease to about 9

percent of municipalities experiencing a 75 percent or higher decrease Additionally a modest

number (n=19 7) of municipalities experienced an increase in their homicide rate

30

Exhibit 8 Percent reduction in homicide rate by number of municipalities

Exhibit 9 presents six negative binomial models for the monthly number of homicides The first

model examines the impact of the gang truce and implementation of the Violence Free

Municipality program and our interaction variable that measures the additive effect of both

the gang truce and the Violence Free Municipalities program on the number of homicides We

found that although the gang truce was associated with a significant decline in homicides the

Violence Free Municipality program was related to a significant increase in homicides13

However countrywide the additive effect of implementing the gang truce and the Violence

Free Municipalities program was associated with a significant decline in homicides

13 To be clear our bivariate analyses showed that the violence free municipalities program was related to a significant decline in homicides For example on the one hand those municipalities that did not participate in the violence free municipalities program experienced a decline in their homicide rate from 406 homicides per 100000 population in the pre-truce period to 321 homicides per 100000 in the post-truce period On the other hand those municipalities that participated in the violence free municipalities program experienced a substantially greater decline in their homicide rate from 752 homicides per 100000 population in the pre-truce period to 392 homicides per 100000 in the post-truce period However after other variables are controlled for in our negative binomial models we found that the violence free municipalities program did not have a positive impact on homicides over and above the gang truce itself

31

Exhibit 9 Negative Binomial models for monthly number of homicides

Model 1 Coefficient (se)

Model 2 Coefficient (se)

Model 3 Coefficient (se)

Model 4 Coefficient (se)

Model 5 Coefficient (se)

Model 6 Coefficient (se)

Homicide - 1 month lag 0010 0010 0010 0010 0010 0010

(0002) (0002) (0002) (0002) (0002) (0002)

Homicide - 2 month lag 0012 0013 0013 0013 0013 0013

(0002) (0002) (0002) (0002) (0002) (0002)

Homicide rate spatial lag 0003 0003 0003 0003 0003 0003

(0001) (0001) (0001) (0001) (0001) (0001)

Gang truce implemented -0483 -0541 -0553 0057 -0544 0564

(0066) (0065) (0064) (0278) (0064) (0279)

Violence free municipality 0497 0183 -0003 -0076 -0037 -0094

(0181) (0167) (0152) (0143) (0153) (0145)

Ln( +1) of MS13 prisoners 0066 0176 0062 0171

(0023) (0028) (0043) (0028)

Ln( +1) of 18th St prisoners 0067 0044 0064 -0043

(0021) (0024) (0021) (0024)

MS13 prisoner spatial lag -0004 0054 -0003 0049

(0045) (0055) (0046) (0056)

18th St prisoner spatial lag 0022 0014 0019 0012

(0027) (0032) (0027) (0032)

Gang truce violence free municipality

-0346 -0294 -0309 -0161 -0304 -0059

(0158) (0156) (0158) (0146) (0158) (0145)

Gang truce Ln( +1) of MS13 prisoners

-0185 -0185

32

(0030) (0030)

Gang truce Ln( +1) of 18th St prisoners

0160 0060

(0025) (0025)

Gang truce Ln( +1) of MS13 prisoners spatial lag

-0139 -0138

(0064) (0064)

Gang truce Ln( +1) of 18th St prisoners spatial lag

0023 0023

(0039) (0039)

Ln( +1) of MS13 on street 0023 0018 0021 0016

(0013) (0012) (0013) (0012)

Ln( +1) of 18th St on street 0043 0039 0040 0037

(0016) (0015) (0016) (0015)

Time 0010 0011 0011 0011 0011 0011

(0002) (0002) (0002) (0002) (0002) (0002)

1month 0064 0065 0067 0067 0067 0067

(0055) (0055) (0055) (0055) (0055) (0055)

2month -0020 -0020 -0022 -0023 -0022 -0023

(0056) (0056) (0056) (0056) (0056) (0056)

3month 0088 0088 0188 0088 0088 0088

(0052) (0052) (0052) (0052) (0052) (0052)

4month -0011 -0011 -0011 -0011 -0011 -0011

(0052) (0052) (0052) (0052) (0052) (0052)

5month 0020 0020 0020 0021 0020 0021

(0052) (0052) (0052) (0052) (0052) (0052)

6month 0000 0000 0000 0000 0000 0000

7month 0036 0036 0036 0036 0036 0036

(0052) (0052) (0052) (0052) (0052) (0052)

33

8month 0047 0047 0047 0046 0047 0046

(0051) (0051) (0051) (0051) (0051) (0051)

9month -0055 -0055 -0057 -0058 -0057 -0058

(0053) (0053) (0053) (0053) (0053) (0053)

10month 0040 0040 0041 0041 0041 0041

(0052) (0052) (0052) (0052) (0052) (0052)

11month 0022 0022 0022 0022 0022 0022

(0052) (0052) (0052) (0052) (0052) (0052)

12month -0040 -0040 -0042 -0043 -0042 -0043

(0053) (0053) (0053) (0053) (0053) (0053)

Urban opportunity factor 0209 0042 0033

(0040) (0040) (0039)

male 10-29 0039 0007 0009

(0021) (0019) (0019)

female-headed household -0006 0002 0000

(0007) (0007) (0006)

unemployed 0001 -0002 -0000

(0006) (0005) (0005)

Racialethnic heterogeniety 0058 -0039 -0087

(0263) (0235) (0230)

Intercept -10571

-10544

-11054 -11618 -11015 -11558

(0064) (0060) (0194) (0241) (0212) (0257)

Ln(Alpha) -2287 -2289 -2293 -2294 -2293 -2294

(0113) (0113) (0114) (0114) (0114) (0114)

Truce random effect 0202 0193 0202 0144 0201 0143

(0039) (0038) (0040) (0031) (0040) (0031)

Intercept random effect 0315 0226 0192 0164 0183 0158

(0047) (0035) (0032) (0027) (0031) (0027)

Cov(Truce and Intercept) -0162 -0099 -0122 -0086 -0114 -0090

(0037) (0031) (0031) (0025) (0031) (0026)

34

The second model includes those variables from Model 1 but also includes variables that

controlled for community-level structural factors In this model the gang truce remains

significantly associated with a decline in homicides but the Violence Free Municipality program

and our interaction variable were no longer related to a reduction in homicide Although the

percentage of residents in a municipality who are male aged 10 to 29 female-headed

households percent unemployed and ethnic heterogeneity were unrelated to changes in

homicide the urban opportunity factor was significantly associated with homicide

Models 3 and 4 included our measures of intervention as well as our measures of the number

of MS13 and 18th Street on-the-street and incarcerated gang members at the municipality level

The analysis showed that while the number of MS13 on the street was unrelated to homicides

the number of 18th Street gang members was associated with an increase in homicide These

analyses also indicated that the number of incarcerated MS13 and 18th Street gang members

from a municipality was associated with a significant increase in homicides We further

examined whether this relationship was associated with the gang truce by including two

interaction variables as presented in Model 4 One measured the interaction between the

number of incarcerated MS13 gang members and the gang truce and a second measured the

interaction between the number of incarcerated 18th Street gang members and the gang truce

The results showed that following the gang truce the number of incarcerated MS13 gang

members was related to a significant decrease in homicides and the number of incarcerated

18th Street gang members was related to a significant increase in homicides

Models 5 and 6 in Exhibit 9 present the same two models (3 and 4) as above but include the

community-level structural variables Model 5 once again shows that the number of MS13

members on the street remains unrelated to the change in homicides and the 18th Street

members on the street was associated with a significant increase in homicide However Model

6 shows that following the gang truce the number of incarcerated MS13 gang members

remained significantly associated with a decline in homicides while the number of incarcerated

18th Street gang members remained significantly associated with an increase in homicides

35

Community-level structural factors were found once again to be unrelated to changes in

homicide rates

Exhibit 10 below further illustrates the relationship between reductions in homicides by

municipality and the municipal-level presence of MS13 and 18th Street gang members in

prison The figure shows that in municipalities with no incarcerated MS13 and 18th Street gang

members no change in homicides occurred following the gang truce However when a

municipality had ten MS13 gang members imprisoned on average that municipality

experienced a 55 percent reduction in homicides When a municipality had fifty MS13 gang

members imprisoned those municipalities on average experienced a 71 percent reduction in

homicide following the truce Conversely our analysis showed that the number of 18th Street

gang members in prison (from a municipality) had a significant and positive impact (that is the

number increased) on homicides following the truce For example if a municipality had ten

18th Street gang members imprisoned on average that municipality experienced a 31 percent

increase in homicides14

14 We examined whether there was an interaction effect between gang members on the street following the gangs truce Analysis for both MS13 and 18th street and their interaction with the gang truce showed no effect

36

Exhibit 10 Impact of the gang truce by number of gang members in prison at the municipal level

The Gang Truce and Crime Displacement

We examined two types of displacement crime type displacement and method displacement

Some policymakers have suggested that although homicides may have declined as a

consequence of the gang truce other forms of criminality or categories of reported crime may

have subsequently increased Crime type displacement occurs when offenders who focus on

one type of crime switch to another form of crime in order to avoid detection or to benefit in

some other way (Eck 1993) In El Salvadorrsquos case some critics of the gang truce have suggested

that as homicides decreased other forms of criminality such as extortion might have

increased substantially due to gang membersrsquo increased freedom to conduct activities inside

and outside of prison (Dudley 2013 Parkinson 2014) Method displacement occurs when

offenders change their tactics or methods of conducting crimes as a consequence of an

intervention (Eck 1993) Following the gang truce in El Salvador some analysts argued that gang

37

members might have begun to hide the bodies of homicide victims to avoid detection and to

ldquoprotect the integrity of the countryrsquos gang trucerdquo (Bargent 2013 1)

As noted above we rely on police data that measured monthly numbers of thefts extortions

robberies rapes and auto theftsrobberies by municipality We also received data on the

annual number of disappearances in each municipality by year Exhibit 11 presents the

descriptive statistics for these variables It shows that the rates of theft and robbery did not

change significantly between the pre- and post-true periods and the rate of extortions

significantly declined The t-tests showed that the rates of rape disappearance and

theftrobbery fromof an auto increased significantly

Exhibit 11 Descriptive statistics for measures of displacement

Mean Std Err [95 Conf Interval]

Theft rate

Pre-truce 1238 049 1142 1335

Post-truce 1213 049 1115 1310

Extortions rate

Pre-truce 353 023 308 398

Post-truce 271 021 229 314

Robbery rate

Pre-truce 544 025 495 593

Post-truce 553 025 504 602

Rape rate

Pre-truce 049 003 043 056

Post-truce 064 005 054 073

TheftRobbery Vehicle rate

Pre-truce 132 022 090 175

Post-truce 165 023 120 210

Disappearance rate

Pre-truce 805 068 670 940

Post-truce 1007 064 882 1132

p lt 05

We further examined the data (excepting disappearance data) similarly to the analyses above

in which we used random effects negative binomial models and regressed the number of

38

crimes (ie theft extortion robbery rape and auto theftrobbery) on various months on

predictors with both temporal and spatial lags Our independent and control variables

remained the same as those used in Model 6 (displayed above in Exhibit 9) We found that

over the study period there was no significant change between the pre-truce and post-truce

periods in the number of thefts extortions robberies rapes and auto theftsrobberies (tables

not shown)

As noted above we also received data on the number of disappearances by municipality and

year but because the data were provided by year there were not enough data points to

examine them temporally Therefore we added this covariate (number of disappearances by

municipality and year) to Model 6 in Exhibit 9 The results of the analysis indicated that the

number of disappearances was not significantly related to change in homicides the gang truce

remained associated with a significant reduction in homicides and our interaction variables

(number of incarcerated MS13 amp 18th Street members truce) remained significantly

associated with homicides

Conclusions

The present study sought to examine the impact of the gang truce on violence and other forms

of criminality We examined homicide data by municipality prior to and after the gang truce

Our outcome variables were obtained from the PNC along with several independent and

control variables obtained from the Ministry of Economy and the National Bureau of Prisons

We analyzed this data using a number of time series and random effects negative binomial

models where we regressed the rate of homicide on various months on predictors with both

temporal and spatial lags and controlled for other contextual factors This type of analysis

allowed us to correct for the number of homicides in one month being related to the number of

homicides in the previous month (ie temporal autocorrelation) Additionally it allowed for

the fact that some communities participated in supplemental interventions (ie violence free

municipalities initiative) which might have impacted the outcome in the same way in a given

month These techniques allowed us to isolate the effects of the gang truce as best possible

We also used data from the PNC that measured by municipality the number of thefts

39

extortions robberies rapes auto theftsrobberies and disappearances to examine the

possibility of the trucersquos impact on crime displacement and diffusion of benefits`

Our findings suggest that El Salvadorrsquos gang truce had a significant and dramatic impact on the

homicide rate The mean number of monthly homicides declined from about 354 prior to the

truce to about 218 following the truce for a net decrease of about 136 homicides per month

Our forecast showed that between March 2012 and June 2014 the truce had saved about 5501

lives As will be discussed further below the analysis suggests that key stakeholders have the

capacity to renegotiate existing norms of violence and that at least some gangs have the

capacity to exert substantial informal social control over their members that can result in

reduced violence

The gang truce also lasted substantially longer than previously evaluated truces Ordog et al

(1995) for example reported that the much publicized gang truce in Los Angeles decreased

homicides by about 35 percent for the first three months but then doubled in months four

through eleven compared with the pre-truce period Similar findings were reported in Trinidad

and Tobago (Maguire Katz and Wilson 2013) where it was determined that homicides

declined for a brief period of time (again for about three months) but then increased

substantially over the long term (12 months) These results suggested that gang truces may

produce short-term benefits yet result in long-term adverse consequences

The findings of the present case study suggests that some gang truces might last longer than

previously believed While the number of homicides began to slowly increase about 12 months

following the truce in El Salvador the results clearly showed that almost two years following

the truce homicides still remained below those experienced prior to the truce

We found however that the gang truce did not result in a homogenous decline in violence

across municipalities About 61 percent of municipalities experienced a decline in homicides

but the decline in violence varied substantially between municipalities For example while

about 16 percent of these municipalities experienced a 25 percent reduction in homicides a

number of others (37) experienced a 75 percent or greater reduction Furthermore it is

important to note that about 20 percent of municipalities experienced a modest increase in

40

homicides This suggests that the impact of a gang truce might be variable and could be

dependent on contextual factors We examined the possible influence of these factors by

assessing the impact of social structural factors and the presence of gangs on the municipal-

level impact of the gang truce While we found that social structural factors were unrelated to a

decline in homicides municipal-level gang presence was associated with the decline in violence

as a consequence of the gang truce

We examined this issue further by parsing out the relative influence of the number of MS13

and 18th Street gang members on the street and in prison from each municipality As noted

above we hypothesized that municipalities where gang member presence was high regardless

of their presence on the street or in prison would experience a greater reduction in homicides

because of their increased influence in these areas We found however that the relationship

was not as direct as we would have suspected In particular our analyses indicated that

following the truce the number of MS13 and 18th Street gang members on the street in a

municipality was not significantly related to a decline in homicide but the number of

imprisoned MS13 and 18th Street gang members was associated with a significant change in

homicides following the gang truce

Of special note was our finding of the differential impact of the truce based on gang Although

we found that the number of imprisoned MS13 gang members was associated with a significant

decline in homicides following the gang truce we also found that the number of imprisoned

18th Street members was associated with a significant increase in homicides following the truce

These divergent findings might be associated with each gangrsquos organizational structure and its

capacity to regulate member behavior

Much prior research suggests that imprisoned MS13 gang members have substantial influence

over violence in Salvadoran communities15perhaps even more so than formal mechanisms of

15 Instituto Universitario de Opinioacuten Puacuteblica (Iudop) La situacioacuten de la seguridad y la justicia 2009-2014 entre expectativas de cambio mano dura militar y treguas pandilleras Aguilar Jeannette (Coord) [et al] --1ordf ed --San Salvador El Salv Instituto Universitario de Opinioacuten Puacuteblica (Iudop) 2014- 260 p graacuteficos tablas cuadros y figuras 28 cm (Talleres graacuteficos UCA) httpwwwucaedusviudopwp-contentuploadslibro_la_situaciC3B3n_de_la_seguridadpdf

41

social control such as the police and courts16 Law enforcement officials for years have claimed

that MS13 is one of the most organizationally sophisticated street gangs in the Western

Hemisphere17 The gang has been widely characterized as having a highly vertical organizational

structure and strong control over criminal enterprises in gang-controlled neighborhoods and as

being decidedly capable of enforcing rules through discipline MS13 leadership resides in the

nationrsquos prison system The Ranfla (gang leadership) is comprised of thirteen MS13 gang

members (PNC 2011) who direct coordinate and authorize street crime and other gang activity

from prison Our findings lend support to the idea that MS13 is more organized than the typical

street gang and that imprisoned MS13 gang members exhibit strong influence over their fellow

gang members on the street

Our findings also suggest however that the gang truce had a boomerang effect in

municipalities with high numbers of imprisoned 18th Street members implying that 18th Street

might not have as much organizational capacity to regulate violence on the streets as MS13

The truce provided incarcerated MS13 and 18th Street gang leaders an opportunity to

negotiate with high-ranking officials and influential diplomats This may have increased their

legitimacy inside and outside of their gangs It appears that MS13 was able to exert its span of

control over the communities in which they had influence and they were able to deliver on the

terms of the gang truce negotiations In the case of 18th Street however incarcerated gang

members may not have had the same organizational capacity for communicating and carrying

out directives In fact a review of the gang truce indicated that there was a conflict taking place

between two fractions within 18th Street Consequently the organizational structure and

culture of 18th Street might be more diffuse than that of MS13 and its leadership structure

might not be as strong because of the internal fractures within the gang This might further

explain why homicides increased in 18th Street communities The internal fractures within the

18th Street gang may have resulted in intra-gang violent conflict that was largely contained

16 Instituto Universitario de Opinioacuten Puacuteblica (Iudop) La situacioacuten de la seguridad y la justicia 2009-2014 entre expectativas de cambio mano dura militar y treguas pandilleras Aguilar Jeannette (Coord) [et al] --1ordf ed --San Salvador El Salv Instituto Universitario de Opinioacuten Puacuteblica (Iudop) 2014- 260 p graacuteficos tablas cuadros y figuras 28 cm (Talleres graacuteficos UCA) httpwwwucaedusviudopwp-contentuploadslibro_la_situaciC3B3n_de_la_seguridadpdf 17 httpwwwlaprensagraficacomeua-declara-a-ms-grupo-delictivo-transnacional

42

within 18th Street controlled territories Further analysis is needed to examine this specific

issue

While not the primary focus of this case study we controlled for any impact that the Violence

Free Municipalities program might have had on homicide in 11 municipalities during the truce

As noted above the Violence Free Municipalities program served as the second phase of the

gang truce (Cawley 2013) The program was first proposed by the two mediators and designed

by the Technical Committee on Violence and Crime Reduction which included representatives

from OAS MJPS the mediators and the Humanitarian Foundation (CISPES 2013) The program

involved the mayors of each municipality collaborating with gang leaders to design prevention

and intervention resources for gang members and at-risk youth The Minister of Justice and

Security David Mungia Payes announced that his ministry would facilitate $74 million in

funding from the OAS UN and other donors to implement the programing In exchange gang

leaders agreed to end violence and other criminality in the Violence Free Municipalities (CISPES

2013) The negotiators also agreed to further discuss both gangsrsquo demand to repeal of the

2010 law that increased the capacity of the police and prosecutors to crack down on gangs

(Ayala 2014)

Our findings suggested however that the Violence Free Municipalities program was unrelated

to change in homicides in these communities While much additional research is needed to

understand why the program was not effective it might have been because the gangs had

already agreed to a truce and had already reduced violence to the extent that they could

Conversely the demands made by the gangs as part of the Violence Free Municipalities

program might have been more than could be delivered The time and resources required to

implement the programming and the political capital that was required to repeal legislation

might have been much more than could be delivered Future research is needed to examine the

processes and impact of the Violence Free Municipalities program

Regardless the base rate of violence in El Salvador was reset for a period of time suggesting

that perhaps the gang truce substantially altered existing norms of violence Klein (1995)

argued that cycles of gang violence (that is perceived or real changes in gang activity) are

43

typically the consequence of seasonality epochal variation (peaks and valleys in violence) and

illusion (it appears as if there is a real change in violence but there is not) In El Salvador we

appear to have observed a self-directed cycle of normative change wherein incarcerated MS13

gang leaders directed a reduction in violence by actuating their organizational span of control

through the gang truce Some of the most powerful and influential gang members in El Salvador

used their political social and economic capital to promote the truce and articulate new norms

of violence They were able to do this by leveraging their informal social control over the streets

through actual or perceived threats of violence against those who violated the terms of the

truce

However following a change in government leadership and the government subsequently

distancing itself from the gang truce the conditions and capacity of MS13 leaders to intervene

in local violence might have deteriorated and violence began to increase substantially This is in

part because the third parties in the negotiation were no longer able to communicate with

government officials about furtherance of the truce Specifically the mediators were no longer

able to negotiate on the behalf of the government and were no longer permitted entrance to

the prisons where they could negotiate with gang leaders

44

Case Study B Gang Trucemdash The Jamaican Experience

In the latest report by the UN Office of Drugs and Crime Jamaica ranked as the sixth most

violent nation in the world with a homicide rate of about 521 per 100000 residents (UNDOC

2011) Policymakers and researchers have attributed the nationrsquos high level of violence to such

factors as drug trafficking (Klein Day and Harriott 2004) access to illegal firearms (Lemard and

Hemenway 2006) and historical processes that include a legacy of conflict between the

nationrsquos two primary political parties (Sives 2002 Figueroa and Sives 2003 Moser and Shrader

1999) all of which have facilitated the entrenchment of the more powerful gangs in

communities of the urban poor Likewise high levels of income inequality and chronic youth

unemployment (Francis and Lyare 2006) problematic urbanization (Stone 1975) social

marginalization and an emergent subculture of violence (Harriott 2008) have contributed to the

nationrsquos violence problem Whatever the causes violence has had a considerable impact on

Jamaicarsquos social and economic development it has decreased investor confidence (Schwab and

Porter 2008) tourism (Harriott 2007) and access to public services It has also increased the

costs related to the health care system (Mansinghand and Ramphal 1993) the criminal justice

system (Caribbean Human Development Report 2012) and the education system (Moser and

Holland 1997)

Jamaicarsquos homicide problem is closely associated with its gang problem The Jamaican

Constabulary Force (JCF) has estimated that some 272 gangs are active in the nation most in or

near the capital city of Kingston (Harriott 2014) Gang types and their respective historical

patterns of conflict matter in Jamaica as these variations may determine their predisposition or

amenability to lasting rather than opportunistic truces Jamaican gangs include territorially

organized crime groups conflict gangs defense crews who regard themselves as defenders of

their communities (Levy 2009) and other less cohesive more transient territorial groups Some

of the latter are predatory others bond around the taken identities that generate conflicts of

other similar groups18

18 These are generally rooted in subcultural issues such as the demand to be treated with respect or with the indiscretions and self-centered aggressiveness of individual members that may be associated for example with sexual competition for the favors of women

45

Estimates of the gang-related homicide rate in Jamaica vary perhaps because the crime is not

clearly defined as far as attributing a death to a gang Regardless researchers agree that the

proportion of homicides that are gang related has increased substantially For example Harriott

(2003) reported a fourfold increase in the rate of gang-related homicides between 1983 and

1997 Likewise Hill (2013) using official police data found an eightfold increase from 2001 to

2009 reportedly in 2001 only about 64 percent (n=73) of the nationrsquos 887 homicides were

gang related compared with 523 percent (n=879) of its 1682 homicides in 2009

To address the problem Jamaica has initiated traditional law enforcement strategies such as

establishing a specialized gang unit (Sinclair 2004) initiating curfews (Sinclair and Tuner 2005)

declaring states of emergency (Jamaican Observer 2010) and implementing community-

oriented policing (Kolpack 2006) It also attempted legislative reforms to curb election fraud

and electoral-related violence that involved local gangs (Levy 2009) None of these strategies

stemmed the tide of gang violence In 2002 the Minister of National Security established the

Peace Management Initiative (PMI) (Henry 2011) to augment governmental and non-

governmental organizational capacity to settle gang disputes in the community through

intervention-based programming such as ceasefires and gang truces

The current study examines the peace initiative instituted by the PMI in Greater August Town

Our objective was to understand the negotiation processes undertaken with and between

gangs and other stakeholders Among other things we were interested in identifying the actors

involved in the negotiations the goal(s) of the negotiations and the strategies employed to

carry them out Most importantly however we wanted to determine whether the gang truce

resulted in the desired outcome a reduction in the number of homicides in the Greater August

Town area

The Greater August Town (Jamaica) Peace Initiative

Greater August Town is located on the northeastern outskirts of the city of Kingston The low-

income area has high rates of youth unemployment and a history of gang-related violence

(Charles 2004 Levy 2009) Over the last decade the Greater August Town community has

sought improved living conditions and revitalization (Levy 2009 95) The arearsquos inherent

46

resilience has been augmented by nearby intellectual and cultural engines such as the

University of West Indies the University of Technology and University Hospital (Charles 2004

38)

Greater August Town is comprised of the communities of August Town proper (which is

fractured into several locales with gang-given names such as ldquoVietnamrdquo and ldquoOpen Landrdquo and

city government-approved names such as Hermitage Goldsmith Villa (Angola) and Bedward

Gardens These socially defined community divisions and subdivisions are markers for the

territorial boundaries of street gangs and therefore in some instances are lines of potential

conflict Some of those boundaries demarcate areas of Greater August Town that are

predominantly supportive of one or another political party but the boundaries do not always

hold political significance politics is but one element in the conflict geography of the area Like

many communities of the urban poor the Greater August Town area is easily mobilized

politically ndash a reality that is understood and at times exploited by street gangs who politicize

gang ldquowarsrdquo in their efforts to build alliances and to neutralize the police In fact the basic

principle of community mobilization in Jamaica is political patronage and clientelism Access to

resources (eg jobs housing education) for the poorest residents is often determined by the

local political party Thus according to Charles (2004 36)

supporters attach themselves to the political parties to get first preference in the

distribution of scarce resources and over time because they are unemployable they

become dependent on their political party for their economic survival These supporters

will kill anyone who threatens the support base of their political party because they

perceive it as a threat to their daily survival

As a consequence some political supporters invest heavily in the electoral contests and

provoke conflicts that affirm their loyalty to their party in order to secure material benefits

from it Political competition is one conflict fault line in what otherwise is a politically

heterogeneous community Specifically political support in Greater August Town is divided

between the Peoplesrsquo National Party (PNP) and the Jamaica Labor Party (JLP) The PNP receives

strong support from residents in August Town proper the upper region of Goldsmith Villa and

47

Bedward Gardens while the JLP is supported by those living in Hermitage and the lower region

of Goldsmith Villa (Charles 2004 ) The division between the political parties in the area as in

much of Jamaica was believed to be at the root of much of the communityrsquos violence

particularly between 1979 and 1993 Some gangs aligned with the PNP and others with the JLP

this often resulted in political boundaries overlapping with gang turf (Charles 2004)

Pre-truce Violence

Nationally the history of gang violence in the Greater August Town area first appeared as

political violence closely associated with the electoral cycle As in many other urban

communities the problem intensified mdash deeply affecting community life in the period just prior

to the national elections of 1980 mdash then continued cyclically until 1993 Gangs have since

harnessed this legacy in order to establish and maintain community support based on common

political affiliations They have used that support to nullify the efforts of law enforcement to

suppress their illicit activities (Harriott 2008)

The conflict profile of these gangs and of the communities in which they are nested has

changed over the decades From the beginning of their involvement in political violence and

territorial control a form of gerrymandering existed that manipulated the voting population

forcing certain individuals out of a particular constituency and preventing those who remained

in the community from voting for the opposing party This was linked both to the electoral cycle

and to the types of inter-gang conflicts that typically were associated with street gangs

(Figuerou Harriott and Satchell) More recently however much of the crime and violence

perpetrated by gangs has resulted from internal conflicts (eg status management disputes

over womengirls or money individual membersrsquo activities that could attract police pressure)

Internal conflict at times has led to gang fragmentation and new alliances that pull more parties

into the conflict escalating the homicide rate and increasing the sense of insecurity among the

general population (Levy 2012)

The most significant of these internal conflicts resulted from the killing of former Jungle 12

leader Neil Wright by members of his own gang Jungle 12 was the most influential gang in

Greater August Town Before his murder in order to increase the gangrsquos access to illicit

48

opportunities in Kingston Wright had been attempting to extend Jungle 12rsquos influence with a

system of alliances with other gangs and recruitment in Kingston (Harriott 2014) In short his

ambition was to transform Jungle 12 from a neighborhood street gang operating on the

outskirts of Kingston into to a dominant organized crime network that could reach into the

heart of the city In pursuit of this goal Wright recruited members from outside August Town

elevating them in the gang hierarchy above the locals This led to status-related conflicts and

resistance to Wrightrsquos leadership within the gang His murder precipitated a split of Jungle 12

into three factions two of them fled to other neighborhoods within August Town (Angola and

Vietnam) resulting in the formation of new alliances and a new conflict geography that

replaced the former political geography of conflict Wrightrsquos killing and the subsequent demise

of Jungle 12 as the dominant gang in Greater August Town altered the balance of power and

escalated inter-gang violence (Harriott 2014) The post-2005 phase of conflict was

characterized by power symmetry conflict intensification and the spread of conflict

throughout the entire geographic area of Greater August Town

Although their origins are unclear retaliatory killings and other violent incidents progressively

intensified between 2005 and 2008 The violence was episodic retaliations were most often

motivated by suspicions related to personal and geographic connections between warring

gangs As the violence escalated new alliances were formed to enhance power and dominance

which in turn increased the number of gangs and gang members involved in the violence

(Harriott 2014) This eventually attracted national attention and triggered community

mobilization for a gang truce

The Truce-making Process

The Greater August Town gang truce was preceded by frequent intense violence and public

outrage As noted above the violence had escalated in November 2005 when Jungle 12 leader

Neil Wright was killed The defection of a Jungle 12 member to Goldsmith Villa (Angola) caused

infighting within the gang and conflict between it and Goldsmith Villa Just a few months later

Wrightrsquos brother Steve and two others were injured during a turf battle (Martin-Wilkens 2006)

49

Thereafter violence began to occur at regular intervals until January 2007 when the Peace

Management Initiative (PMI) hosted a peace march in the community Two PNP politicians

urged the community to unite A PMI leader declared that the peace march was being held to

ldquodemonstrate to the public that Jungle 12 members are back together and that they want

peacerdquo (Thompson 2007 1)

Although hopeful some residents remained skeptical about the peace march perceiving the

action to be politically motivated In the absence of trustworthy information inter-group

conflicts tended to be interpreted through a politically partisan lens this created obstacles to

isolating the gangs building a consensus for peace and unifying community mobilization The

politically based narratives weakened the communityrsquos leverage for peace as well as the

exposure of the gangs to police action As one resident said ldquoThe election is coming up and

they want[ed] the people to vote for the PNP is one of the main reasons why they have to walk

todayrdquo (Thompson 2007 1) Those who shared such views stayed away from the peace march

Although that widely held myth was not factual it did serve to demoralize and demobilize one

part of the community A local UWI faculty member was articulate in his assessment of the

politics behind the march

August Town violence is not violence of organized crime which is based on drug

trafficking extortion or some other criminal enterprise [but] rather the violence in

August Town is essentially ldquotribalrdquo mdash the Peoplersquos National Party tribe versus the

Jamaica Labour Party tribe [which has been complicated by a splintering within the

PNP tribe] (2007 1)

Indeed it was not the violence of organized crime but neither was it political violence The

individual quoted above neglected to mention that the conflict was between those pro-PNP

splinter groups who were largely comprised of members of Jungle 12 Moreover their ldquopro-

PNP-nessrdquo was unrelated to the conflict there was no factional infighting within the local PNP

organization at that time

Nevertheless following the peace march the gang violence diminished Then in November

2007 a turf war erupted between two gangs from the Greater August Town neighborhoods of

50

Vietnam and River This time as the police stated the gang violence was less about politics and

more about dominance and turf Police were dispatched to perform directed patrols but

whenever they were not present the shootings continued (Mcleod 2007) In April 2008 the

community witnessed local gangs engaging in a five-hour-long street battle that left two killed

and three others wounded It ended only after the police deployed armored trucks The next

month another round of gang violence resulted in five others being killed including a one-year-

old child This resulted in the three members associated with the gang who committed the

homicides being killed in retaliation (Virtue 2008) The local community mobilized against the

violence increasingly cooperating with the police providing more information about the gangs

Subsequently gang members observed a decline in their influence within the community

During the early period characterized by low-intensity conflict the less influential gangs at

times used manipulation of the police as a tactic for suppressing the more influential gangs

This was largely done through strategic release of information Prior to 2005 when Jungle 12

was dominant its membersrsquo illicit activities were constantly reported to police by members of

other gangs as a means of compelling a compromise or settlement of conflicts In practice this

was done by ldquotrading casesrdquo Once a crime had been investigated by the police and suspects

had been charged an opportunity was created for the gangs and other parties to the conflict to

settle the matter by agreeing to drop their cases (typically by ceasing cooperation with police

investigators) This type of ldquoself-helprdquo served to end some of the retaliations but it rested upon

the manipulation of the police (Harriott 2014)

Later in an attempt to quell escalating inter-gang violence the police established buffer zones

between the warring gangs This action resulted in unintended consequences For example

when the police declared a buffer zone between August Town and Hermitage Hermitage took

advantage of the opportunity to attack Angola Some Angola residents accused the police of

turning a blind eye and creating an opportunity for Hermitage to attack their community

Although little reliable information exists about why the police made the deployment the way

that they did it is more likely that the police inadequately assessed the situation (ie mis-

assessed the pattern of alliances and the likely targets of attack) (Harriott 2014) In the areas

affected by this kind of increasing violence community members became angered and lost

51

confidence in the police The error resulted in some parties to the conflict receiving increased

support from their communities and in greater gang-community cohesion (Harriott 2014)

After a brief period the police identified this problem and began to disengage by no longer

providing a buffer between gang controlled areas which in turn allowed still more conflict to

occur between the gangs

As the violence escalated beyond their control police finally responded by applying their own

forms of pressure For example units under the direction of the JCF High Command would

make periodic raids in the community during which they would at times seize weapons and

make mass arrests (Sinclair 2005) However there were also moments when the local police

were very responsive improved their relations with the community and consequently gained

greater access to relevant information Two such moments occurred just prior to and again

immediately after the truce moments during which there was greater freedom of movement

and open collaboration between the community and the police (Harriott 2014)19

The Establishment of the Greater August Town Gang Truce

The Greater August Town gang truce was led by the Jamaican Peace Management Initiative

(PMI) The PMI is a government-funded initiative created for the purpose of working with gang

members to reduce violence Due to community mistrust of the police in 2002 the organization

was established as an alternative organizational mechanism for responding to gang violence

The PMI sought to bridge government and civil society efforts to mediate disputes between

gangs as well as to provide outreach to gang members (Bakrania 2013) While efforts to

institute a gang truce in Greater August Town were led by the PMI a number of other

stakeholders helped to facilitate the truce these included faculty at the University of the West

Indies (UWI) and representatives from the police the faith-based community and the August

Town Sports and Community Development Foundation (Jackson 2008 Levy 2009 also see

Appendix C) The gangs involved in the truce included those from August Town Hermitage

19 The quality of police-community relations largely depended on the style of the local station commander however regardless of the external environment

52

Goldsmith Villa Bedward Gardens and African Gardens Because of its formality as well as its

perceived effectiveness the truce signed on June 24 2008 was regarded by many as the first

of its kind in Jamaica (Levy 2009)

Truce negotiations began early in June 2008 and lasted for about three weeks The gangs

sought to leverage their violence-making capabilities and demanded payment for peace They

asked the third-party negotiators for money ldquoworkrdquo and start-up funds for proposed micro-

businesses (Wilson 2014) Those demands were rejected by the negotiators on the grounds that

the third-party institutions would not buy a peace that was intended to save the lives of those

who were making the demands Moreover if peace was to be purchased then gang conflict

could be used continuously to extract money and other benefits from negotiators The third-

party actors made some demands of their own In some quarters of the community and society

the surrender of guns was viewed as a litmus test of the sincerity of the gangs Consistent with

this the negotiators suggested that the parties to the conflict symbolically hand over one gun

each that suggestion was immediately rejected by the gang leaders These kinds of demands

from the various parties ceased after a time as they all agreed that the truce was to stand on

its own merits (Harriott 2014)

As the truce began to be committed to paper a number of stakeholders expressed concern that

their greatest risk in participating could be the potential for Jungle 12 factions to use the peace

agreement as they had in the past as a tactic to persuade their enemies to let their guard

down Others however recognized that Jungle 12 had now been weakened and that a formal

public peace agreement would be beneficial to the gang and therefore this time would be

different (Harriott 2014)

The gangs held fast to their claim that their weapons were needed for their own protection

because the police were ineffective in responding to violence in their communities (Jamaican

Gleaner 2014) It became a precondition of the truce that the gangs would not be required to

turn in their guns and other weapons (Jackson 2008) The truce agreement did specify

however that ldquoall persons are allowed to move freely across all boundaries regardless of

reputation or affiliation No gun salute or any other shooting is to take place in the community

53

for a period of at least five yearsrdquo (2008 also see Appendix C) The truce agreement and its

conditions were prescribed in a document that was finally signed by all of the major

stakeholders including the gangs (see Appendix C)

Throughout the negotiations each of the gang leaders had attempted numerous times to use

the truce as an opportunity to bargain for money jobs and business support grants Such

demands consistently were rejected by the third-party actors Nonetheless both prior to and

after the truce some efforts were made to create better opportunities for young people

residing in the community UWI for example provided a homework supervision program to

encourage students to further their education and it developed a community-building initiative

to help improve schools and enhance sports programming (Levy 2009) Such programs were

conducted as part of UWIrsquos Township Project in August Town which invested significant

resources in developing the residentsrsquo job-related capabilities and collective self-efficacy

The Greater August Town gang truce was noteworthy for two reasons First the gang truce

received substantial press attention The media were invited to witness the ldquosigningrdquo of the

truce by the gang leaders in the presence of a JCF Deputy Commissioner of Police a PMI board

member and the UWI Principal and two professors of its faculty Second the truce was widely

credited with decreasing violence in Greater August Town and it served as an exemplar to

other communities seeking to replicate its success (Virtue 2008) A number of reports

manuscripts and newspaper articles proclaimed the truce to be a success Bakrania (2013 10)

for example reported that ldquoPMI has been credited with stopping gang wars in August Town

rdquo Levy (2009 94) remarked that the ldquomost interesting outcome of PMI efforts to date was the

Peace Agreement reached in August Town in late 2008rdquo Likewise a government report noted

that ldquothe peace treaty was a pivotal achievement in August Town that has significant potential

for wider application Crime levels dropped markedly in August Town after the signing of the

peace agreement in June 2008rdquo (McLean and Blake-Lobban 2009 78) To this day August Town

celebrates the signing of the truce with an annual celebration with food and music

(Cunningham 2011)

54

Methods

Our evaluation relied on a pre-testpost-test quasi-experimental group design Our

methodology examines the Greater August Town community which is comprised of three

contiguous towns where the gang truce took place (the target area) and the balance of Jamaica

which is comprised of 178 communities (comparison areas) As seen in Exhibit 12 the average

number of residents living in each of the three communities in the target area was not

significantly different than that for the rest of Jamaica about 7776 residents lived in each of

the Greater August Town communities compared with 6468 in the other communities

Likewise communities of Greater August Town were about as densely populated as other

communities (2960 per square kilometer versus 2647 per square kilometer) and the age

range of residents was similar as well However Greater August Town (a) had a significantly

higher proportion of its residents living in poverty (196 vs 158) (b) consumed fewer

resources than other communities and (c) reported significantly more homicides than other

communities prior to the truce (see Exhibit 12)

55

Exhibit 12 Descriptive characteristics of Greater August Town and balance of Jamaica (2007-2011)

Comparison

Area Greater August

Town All areas

Population (mean) 6468 777633 648994

(sd) 720482 353731 715621

Population density (mean) 264719 296033 265238

(sd) 271023 285501 270465

Percent in poverty 1577 1957 1583

(sd) 1036 106 1029

Consumption 15737890 1106939 1566048

(sd) 10713020 205336 1064021

residents under 15 yrs old 2369 2494 2371

(sd) 487 115 484

residents 15-65 yrs old 6840 6901 6841

(sd) 423 29 419

Murder per month (mean) 674 857 677

sd 1928 1409 1920

Total murders 10068 180 10248

n 178 3 181

plt=05

Measures

Two distinct data sets were merged to measure the impact of the Greater August Town truce

First data from the 2011 decennial census provided community-level measures of the social

and economic characteristics of the 181 communities in Jamaica Described in detail below the

community-level data used in the study included population population density gender age

poverty and consumption20 These data were obtained directly from the Statistical Institute of

Jamaica

Second police homicide data from the years 2007 through 2011 were used to construct the

studyrsquos community-level measure of homicide The homicide data were aggregated by month

20 Consumption is an alternative measure of poverty in Jamaica which measures the consumption of food and non-food items

56

and appended to the community-level data The final (merged) data set included 10248

homicides over the 60-month study period These data were obtained from the Jamaica

Constabulary Force (JCF)

The dependent variable examined in the study was constructed from official police homicide

data Once again the homicide data represented the number of officially recognized homicides

in Greater August Town and each of the remaining communities in Jamaica We examined

change by comparing the homicide data 18 months prior to the truce with the homicide data 42

months following the truce More specifically we examined whether there was a change in the

number of homicides in the 30 days following the truce (month 1) as well as whether the truce

had an impact every three months thereafter (ie months 2-5 6-8 9-11 12-14 15-42) and

whether any changes in homicide coincided with changes in homicide in the balance of

observation areas The frequency distribution of our dependent variable is presented in Exhibit

13 It shows that prior to the truce the target area on average experienced significantly more

homicides (1495) than did the comparison areas (920)

57

Exhibit 13 Distribution of homicides in the target and comparison areas

Comparison

Area Target

Area Total

Pre-truce period Mean 920 1495 932

SD 2469 1966 2461

N 241400 5100 246500

Month 1 of truce Mean 741 286 733

SD 1785 496 1772

N 17800 300 18100

Months 2 thru 5 of truce Mean 647 905 652

SD 1821 1249 1812

N 71200 1200 72400

Months 6 thru 8 of truce Mean 577 1236 588

SD 1690 1074 1683

N 53400 900 54300

Months 9 through 11 of truce Mean 718 333 711

SD 2034 733 2019

N 53400 900 54300

Months 12 through 14 of the truce Mean 687 095 678

SD 1519 286 1509

N 53400 900 54300

Months 15 thru 42 of the truce Mean 564 589 564

SD 1683 1042 1674

N 516200 8700 524900

Total Mean 674 857 677

SD 1928 1409 1920

N 1006800 18000 1024800

An illustration of the trends in homicide prior to and following the gang truce are shown in

Exhibit 14 It shows that 30 days following the truce homicides fell in the target and

comparison areas then increased and decreased several times with a general downward slope

in violence over time

58

Exhibit 14 Monthly number of homicides pre-post truce in the target and comparison areas

We also used a number of measures to control for community-level structure from the 2011

decennial census These community-level data included the communityrsquos population

population density (per square kilometer) and community level of consumption Additionally

the census data included measures of the percentage of the population that was female under

15 years old 15 and 65 years old and 65 years old and older as well as a measure of the

percentage of the population living in poverty Principal components analysis was used to

reduce some of these data into a summary measure

Exhibit 15 shows the results of the component loadings One component was extracted that we

designated as socio-economic status (SES) which exhibited high loadings for percent living in

poverty percent under 15 years old percent 15 to 65 years old and consumption Excluded

from the principal components analysis were population and population density Population

was used as our exposure variable and population density was logged to address skewness in

these data

Exhibit 15 Factor loadings from principal components factor analysis

Loading Poverty 78 Consumption -76 under 15 years old 92 between 15 and 65 years old -80

59

Analytic Strategy

In order to test whether the truce had an impact on homicides in the target area andor

whether displacement had occurred in the balance of the study area several analytic

techniques were employed Most of the methods employed the use of the homicide rate as the

dependent variable We explored the data in this way to provide the maximum statistical

power to detect an effect As a check on these methods we also employed a generalized model

to compensate for the non-normality of our outcomes

First focusing only in the target area we performed a simple t-test comparing the homicide

rates before and after the truce (the unit of analysis was a month) however this technique had

limitations the most severe of which was that even if the test were significant it would be

difficult to determine whether the difference was due to the gang truce or a natural change

over time in the outcome Second to address this limitation time series models were

employed whereby the homicide rate for the target area was modeled as a function of time

with truce period indicators included to measure the effect of the truce net of the temporal

trends These models were estimated with ARIMA techniques with a one-month lag auto-

correlated error Third we examined the homicide rate for each town using a panel time series

model In this model the temporal trend for each town was examined with indicators for target

areas and truce periods included The main effects for the truce periods measured the effect of

the truce in the target areas and the moderators of the truce period in the comparison areas

measured displacement effects Finally because the dependent variable coded is not normally

distributed across months we used a negative binomial time series model to estimate the

number of homicides with the population covariate serving as an exposure variable

Findings

The first set of results examines only the target area The first test was a simple t-test

comparing the mean homicide rates before and after the truce periods The result was a mean

difference in the homicide rate of -890 per 100000 with a significant t-statistic of 370 While

60

this result is statistically significant we caution that it may or may not reflect an impact of the

truce To further examine the truce effect in the target area we performed ARIMA regressions

The first model did not include an effect of the temporal trend In this model we again find that

by month 15 the murder rate decreased by about -89 per 100000 (Exhibit 16)

Exhibit 16 Results of basic ARIMA model

Next we employed the ARIMA model again but included a variable (date) to control for the

temporal trends in the data Exhibit 17 shows that when we controlled for temporal trends the

impact of the truce we observed was no longer significant This result indicates that it was not

the truce per se that caused the decline in homicides but instead the decline in homicides was

part of a larger (local and nationwide) decline in homicides

Note The test of the variance against zero is one sided and the two-sided sigma 7877339 5907266 1334 0000 6719536 9035142 L1 1720758 1242218 139 0166 -0713944 415546 ar ARMA _cons 1538436 1904285 808 0000 1165203 1911669 t_15 -8888017 2969383 -299 0003 -147079 -3068134 t_12 -1507749 1820406 -083 0408 -507568 2060181 t_9 -132475 9189969 -144 0149 -3125951 4764507 t_6 -2861008 6335492 -045 0652 -1527835 9556328 t_2 -5166363 4498142 -115 0251 -1398256 3649834 t_1 -1111771 4463623 -025 0803 -9860311 7636769murders_rate murders_rate Coef Std Err z Pgt|z| [95 Conf Interval] OPG

Log likelihood = -2089969 Prob gt chi2 = 00397 Wald chi2(7) = 1472Sample 564 - 623 Number of obs = 60

ARIMA regression

61

Exhibit 17 Results of ARIMA model with control of temporal trends

We next estimated the possible displacement effects of the truce Exhibit 18 presents the

results of these models Examination of the main effects of the truce period does not indicate

any effects and looking at the truceComparison interaction effects we also do not find any

displacement effects Note that these models also controlled for the socio-demographic

characteristics of each community

Note The test of the variance against zero is one sided and the two-sided sigma 7775398 6622814 1174 0000 647735 9073446 L1 1270618 1253989 101 0311 -1187156 3728392 ar ARMA _cons 1480274 1197324 124 0216 -8664378 3826986 date -2318963 2091772 -111 0268 -6418761 1780835 t_15 -3519687 7694691 -005 0964 -1543329 1472935 t_12 -970747 2228007 -044 0663 -5337561 3396067 t_9 -8844212 9402792 -094 0347 -2727335 9584921 t_6 2995697 6576644 005 0964 -1259042 1318956 t_2 -2353533 4923798 -048 0633 -12004 7296933 t_1 -9194625 5462116 -017 0866 -1162501 9786088murders_rate murders_rate Coef Std Err z Pgt|z| [95 Conf Interval] OPG

Log likelihood = -2082031 Prob gt chi2 = 00710 Wald chi2(8) = 1444Sample 564 - 623 Number of obs = 60

ARIMA regression

62

Exhibit 18 Results of Panel (Town) time series model with control of temporal trends

Last we used a random effects negative binomial regression that predicted the homicide rate

with population as an exposure variable and controlled for the socio-demographic

characteristics of each community The results are presented in Exhibit 19 The analysis showed

that time had a negative effect indicating that homicides were decreasing in all areas over the

study period The main effects of the truce (truce = 1 2 ) represented the effects of the

truce in the targeted area and did not show a significant effect for any period following the

gang truce However we did find that the homicide rate significantly increased in the

_cons 6310151 3518738 179 0073 -5864483 1320675lnpopulationdensity 121961 1709831 713 0000 8844891 1554731 ses 1245334 2428629 513 0000 7693312 1721336 date -1276623 0345659 -369 0000 -1954101 -0599144 15Balance 5392222 4128863 131 0192 -2700202 1348464 12Balance 1318516 7903159 167 0095 -2304746 2867507 9Balance 1042658 7925876 132 0188 -5107854 2596101 6Balance -1703928 7919006 -022 0830 -1722489 1381704 2Balance 3498997 7175155 049 0626 -1056405 1756204 1Balance 1056874 1104696 096 0339 -1108291 3222039 truceaugust Balance -4757021 3276618 -145 0147 -1117907 1665032 august 15 -4269255 4284086 -100 0319 -1266591 41274 12 -1262496 7871907 -160 0109 -2805361 2803697 9 -1014276 7881965 -129 0198 -2559113 5305603 6 6563019 786731 008 0934 -1476334 1607595 2 -5034368 7123478 -071 0480 -1899613 8927392 1 -1215158 1095699 -111 0267 -3362688 9323715 truce murders_rate Coef Std Err z Pgt|z| [95 Conf Interval]

Prob gt chi2 = 00000 Wald chi2(16) = 12359 max = 60 avg = 5661878Estimated coefficients = 17 Obs per group min = 48Estimated autocorrelations = 1 Number of groups = 181Estimated covariances = 1 Number of obs = 10248

Correlation common AR(1) coefficient for all panels (02045)Panels homoskedasticCoefficients generalized least squares

Cross-sectional time-series FGLS regression

63

comparison areas in months 12 through 14 following the truce In particular we found a 29

percent increase in the homicide rate in the comparison communities for that period (exp (-

1797 + 2048) = 1285 p-value = 004) Since this effect is only significant at the 005 level

however and given the number of analyses used to examine the data it is possible that we

found this effect by chance alone

64

Exhibit 19 Random Effects Negative Binomial

65

Given these caveats we visualized this model with the following set of marginal predictions as

observed in Exhibit 20 We saw that the targeted area (as illustrated in red) experienced an

immediate decrease in homicide which coincided with an increase in homicides in the balance

of the study area However the target area quickly returned to ldquonormalrdquo and homicides in the

comparison area decreased again During months 9 through 11 following the truce there was a

reduction in homicides in the target area with an associated increase in the comparison area It

is important to point out that the confidence intervals are large and we cannot yield concrete

conclusions from these results However it appears that the truce might have had a temporary

short-lived displacement effect decreasing homicides in the target area but increasing

homicides in the comparison area

Exhibit 20 Predicted change in homicides in the target and comparison areas

Conclusions

From 2000 through 2009 Jamaica experienced a substantial number of homicides many of

which were attributed to gangs in one form or another Traditional law enforcement responses

were repeatedly implemented but until 2010 those had little effect Some policymakers in

Jamaica as well as in other nations throughout the Caribbean and Central America have

recently been experimenting with novel approaches to reducing gang-related violence notably

the implementation of gang truces In Jamaica at least eight gang truces reportedly have been

66

negotiated since 2001 (Levy 2009) The Greater August Town gang truce was thought to have

been one of the more successful and it has served as a model for other communities to use

(2009) Our purpose here has been to identify the actors involved in the negotiations of that

truce the negotiation goals and the implementation methods used and then to examine

empirically the impact of that truce on homicide rates in the targeted community

The 2008 gang truce in August Town was a response to violence that arose when the leader of

one gang was killed creating a power vacuum that other gangs saw as an opportunity to

increase their influence in the community Concomitantly the community as well as the gangs

feared that an absence of formal social control would result in further violence The police

reacted unevenly At some times they engaged in appropriate but heightened levels of

preventive patrol while at other times they purposely provided little or no protection on

occasion they used aggressive tactics that further isolated them from the community The end

result was that there was neither stability nor predictability in the police response and

therefore little trust in the police to address the problem

As the violence further escalated the community mobilized The Jamaican Peace Management

Initiative faculty members from the University of the West Indies (UWI) the Jamaican

Constabulary Force (JFC) and other community-oriented groups joined forces seeking to

reduce the increasing number of homicides by brokering a truce between the gangs Over the

three-week negotiation period the negotiators and the gangs sought terms from one another

The gangs wanted payment ldquoworkrdquo and funds for micro-business development to end the

violence The third-party stakeholders wanted the gangs to disarm actually or symbolically

Neither the gangs nor the stakeholders had substantial leverage nor did they have much to

offer one another in terms of incentives In the end however a truce was agreed upon and all

of the gangsrsquo leaders and several key community stakeholders signed it at a public ceremony

with the media in attendance

At first glance our impact findings appeared to show that the gang truce was an effective

mechanism for reducing violence Bivariate analyses showed a significant decline in homicides

after the truce was implemented This explained the work previously published by

67

policymakers researchers and news reporters Upon further examination of the data however

comparing change in the target and comparison areas and accounting for temporal trends we

found that the decline in homicide was part of a larger nationwide decline in violence and that

the gang truce was not responsible for the decline The only significant effect that we

uncovered was that possibly the homicides were displaced outside the target area for a brief

period of time but then returned to normal

Any one of a number of explanations might be offered for the strategyrsquos lack of effectiveness It

might be that the Jamaican gang leaders at least those in Greater August Town did not have

the organizational capacity to change gang member behavior Much prior research suggests

that in general gangs have limited organizational structure and little formal leadership This

might suggest that gangs do not possess the necessary capacity to regulate their membersrsquo

violence That said gangs in Jamaica including in Greater August Town have been found to be

fairly organizationally sophisticated and to possess strong leadership

In fact in a small number of Jamaican communities gangs have been found to be highly

organized with individual gang leaders being referred to as dons and community leaders The

gang leader in such a community is often found to have substantial control over members and

residents as these communities often turn to the don rather than the police for justice The

don will hold court and punish those who commit crime Punishment can include beatings and

torture as well as execution (Morgensen 2004) Although this level of organizational structure

and sophistication is found only in a small number of Jamaican communities generally the

gangs in Jamaica are believed to have some organizational capacity or at least enough to

reduce violence in communities

Our findings however indicated that prior to 2005 and the death of Neil Wright perhaps only

Jungle 12 could approximate that capacity to discipline members and enforce a truce After the

gangrsquos fragmentation in 2005 Jungle 12 lost much of its organizational capability and

enforcement of the truce was therefore difficult The truce negotiators sought to address the

enforcement issue by proposing a peace council that would involve all parties The proposal

was approved by all key stakeholders still some gang leaders demanded cash payments as a

68

condition for attending council meetings Peace was consistently seen by then as a bargaining

tool rather than as an honest attempt to establish and maintain peace In the end members of

only two gangs were attending the meetings21 and the council soon dissolved

In an effort to replicate the council function UWI sponsored one of the most respected

negotiators a community activist to become a one-person monitoring and intervention

specialist or a ldquoviolence interrupterrdquo His job was to ensure that truce violations did not lead to

a return of the gang wars mdash and there were many violations of the truce For example there

were instances of gang members crossing boundaries and entering the turf of another gang

armed although not initiating conflict behavior that was interpreted by the opposing gangs as

preparation for the next round of ldquowarrdquo or as laying a foundation for a surprise attack that

would exploit the truce for this purpose In the absence of the council these matters were

reported to the violence interrupter who tried to resolve the problems in consultation with the

various gang leaders Often the gang leaders were unresponsive or incapable and therefore the

threatening practices and violence continued Ultimately there were no rules or bodies or

persons who could regulate the violence and there were never any reference points for

compliance The formal truce agreement was an attempt to negotiate and impose such rules

via a collective pressure that would include third parties but it was unsuccessful in doing so

The potential for re-engineering norms related to conflict thus was not realized

Another explanation for the failure of the gang truce might be that it was more a vehicle for

rhetoric rather than for reality The gang leaders insisted that they would sign the truce

agreement only if it were ratified in public with the presence of the media (Jackson 2008 Levy

2009) The leaders might have viewed the process in and of itself as a means of increasing their

reputation and influence within the community and in policymaking circles (and to reduce

mutual distrust) In signing the truce gang leaders publicly pledged to reduce their involvement

in violence thereby calming local residentsrsquo fears They also made public efforts to increase

resources for their communities perhaps in an attempt to portray themselves as ldquoprovidersrdquo to

the community In fact the truce did provide gang leaders with an opportunity to be seen in

21 Interestingly the Jungle 12 factions did not attend any of the peace council meetings

69

public collaborating with important community stakeholders The imagery of the public signing

was of the government (via the PMI) and others approaching the gang to ask them to use their

means of informal social control in the community to reduce violence mdash to accomplish

something that the government could not do on its own As a consequence the process may

have been perceived by gang leaders as a victory because it enhanced the gangsrsquo reputation

with both the government and community

Alternatively from the start the gangs might not have been fully invested in the gang truce

One of the major criticisms of the Greater August Town gang truce was that gangs were not

required to give up their firearms although some believed that this was an unrealistic request

their demand and the demand of many that all guns be turned in immediately was

quite unrealistic given the decades of ingrained gun culture and the continued inability

of the security forces to guarantee protection for any corner against armed rivals It was

obvious to most observers that that kind of situation could not be ended overnight and

that this was a reasonable first step in the process (Levy 2009 63)

The gangs feared that if they were to disarm themselves they would be vulnerable to other

gangs and unable to protect themselves a concern that appears not to have been addressed by

mediators Indeed at times some elements within the community felt somewhat dependent on

the gangs to maintain security If the gangs would have been disarmed and there were no

near-term alternative prospects for any form of social control both the gang and the

community might have faced additional violence as has been observed in the past In the end

the gang truce only called for a reduction in gang violence and did not provide any solutions to

address the larger problems between the gangs nor did it provide the gangs with any tangible

benefits for abiding by the truce

70

Case Study C Gang Trucemdash The Honduran Experience

Introduction

Violence in Honduras is at epidemic levels increasing almost 44 percent over the past five

years In 2012 there were 7172 homicides in Honduras or about 86 homicides per 100000

population (Instituto Universitario de Democracia Paz y Seguridad 2013) making it the most

violent nation in the world (United Nations 2013) Likewise Hondurasrsquo second largest city San

Pedro Sula has the highest municipal level homicide rate in the world with 1290 homicides

(Instituto Universitario de Democracia Paz y Seguridad 2013) or about 174 per 100000

population (United Nations 2013) In comparison the average homicide rate across the globe is

about 62 per 100000 and the average homicide rate in Central America is about 27 per

100000 (United Nations 2013)22

Much of the discussion about the causes of Hondurasrsquo high homicide rate has focused on its

relationship with international drug trafficking routes gangs and conflict between crime

groups and the government and government instability Estimates of gang involvement very

widely but some have suggested that there are between 12000 (Seelke 2012) to 36000

(Ratcliffe et al 2014) gang memberrsquos in Honduras who typically belong to one of two gangs

MS-13 and 18th Street These gangs are said to be less organized than their counter parts in El

Salvador but are said to be just as involved in extortion and intimidation and perhaps more

involved in drug trafficking because of their stronger linkages with Mexican drug cartels

(Wilkinson 2013)

22 An unusual characteristic of the homicide problem in Honduras is the age of victims Typically in the Western Hemisphere homicide victims are aged 15 to 29 In Honduras however those 30 to 44 have the highest rate of violent victimization For example 1 out of 280 males 30 to 44 years old are the victim of homicide compared to 1 out of 360 males 15 to 29 years old (United Nations 2013) These findings by themselves are suggestive of a chronic gang problem (Spergel 1995) Honduras also stands out in the Western Hemisphere in the proportion of its homicides that involve a firearm In 2012 about 84 percent of the homicides involved a firearm (11) The proportion of homicides that involve a firearm appears to be increasing as well In 2008 79 of homicides involved a firearm compared to 81 in 2009 83 in 2010 (United Nations 2013)

71

Over the past decade the nation has responded with ldquoMano Durardquo (ie iron fist or heavy hand)

The new legislation provided the police with more authority to stop search and detain gang

members The new legislation also permitted the courts to sentence gang members to prison

for 12 years for simply being a member of a gang and allowed the courts to sentence

individuals to even longer prison terms for gang related incidents Concomitantly the military

joined the effort to fight gangs by patrolling neighborhoods along side the police While the

public and media strongly supported the shift in national policy toward Mano Dura much of

the evidence suggests that these legislative and policy changes were not effective as the

number of homicides continued to escalate Some suggest that its lack of success was because

gang members who were arrested were released due to of lack of evidence or those who went

to prison if they were not a gang member before entering prison joined a gang Other critics

point out that the heavy handed approach by the government led to loss in the rule of law as

vigilantes engaged in extra-judicial violence against gang members (Seelke 2012) Still others

said that the legislation and policies never really had a chance of working because of the

general lack of effectiveness of the police and courts and the wide spread corruption

throughout the criminal justice system (Zilberg 2011)

As a consequence of the above policymakers and citizens voiced optimism about the possibility

of a truce between gangs after initial results in El Salvador suggested the strategy might be

effective (Villiers-Negroponte 2013) Honduran church leaders and the Organization of

American States (OAS) began to develop a strategy to implement a similar type of truce in

Honduras and the President offered his personal support in their efforts (Arce 2013) In this

case study we examine the processes that lead to the Honduras gang truce and the nationwide

impact of the truce on homicides In the below section we discuss the major stakeholders who

participated in the truce processes leading up to the truce and the establishment of the truce

Key stakeholders

The primary facilitator for the truce process in Honduras was Archbishop Roacutemulo Emiliani who

had earlier served as the Assistant Bishop of the Dioceses of San Pedro Sula and who received

72

support from the Catholic Church to pursue the truce (Bosworth 2013) Prior to the

negotiations he was well known for his work which attempted to establish peace between the

gangs and his advocacy for prison reform and social reintegration programs for gang members

(The Daily Herald 2013) From the onset Monsignor Emiliani proceeded cautiously to ensure

reasonable expectations among the public and policymakers He maintained publically that ldquohe

didnacutet want to be a salesman of false promises about what was going to occur in the future the

things that they do are unpredictable but we expect to have a declaration of reconciliation

principles with societyrdquo (El Mundo 2013) Additionally he wanted to set reasonable

expectations because he knew that it would be a ldquoslow painful and draining processrdquo (El Nuevo

Siglo 2013) and that ldquoWhat is coming is difficult It is not easy It is complicatedrdquo (Castillo

2013)

As in El Salvador the Organization of American States (OAS) played a major role in facilitating

the peace process alongside Monsignor Emiliani Adam Blackwell served as the Secretary of

Multidimensional Security for the OAS and represented Canada on the Honduras Security

Reform Commission (Willcocks 2014) His participation in the mediation process was requested

by Honduran gang members who were in prison They requested that the OAS help broker a

peace agreement with the Honduran government and to help identify resources that would

assist gang members to obtain legitimate jobs (Associated Press 2013) The OAS together with

the Catholic Church served as a ldquobridgerdquo between the executive branch of the government and

the two gangs Additionally two of the mediators (ie Salvadoran Army officer and Police

Chaplain Monsigor Colindres and former Salvadorian congressman Mijango) who helped broker

the truce in El Salvador provided additional support to Monsignor Emiliani and Secretary

Blackwell They traveled to Honduras to present their experiences with the gang truce in El

Salvador and to convey that a gang truce is a promising and legitimate strategy for addressing

gang violence (Associated Press 2013)

Gang leaders of the two primary gangs in Honduras (MS 13 and 18th Street) also participated

extensively in the negotiation process It was stated that they had become weary of the violent

conflict and understood that a truce would be beneficial to the Honduran people (Servellon

73

2013) From the beginning however a number of the critics of the truce argued that Honduran

gangs did not have the capacity to control street level violence They characterized the

Honduran gangs as having less organizational leadership (Bossworth 2013) less control over

turf (LatinNews Daily Report 2013) and being more organizationally fractured (Dudley 2013)

than MS 13 in El Salvador

At the time that discussions about the possibility of a gang truce began President Porfirio Lobo

Sosa was publically supportive of the Catholic Church and OAS negotiating with the gangs

Media reports quoted the President saying I am ldquoprepared to do what ever is necessaryrdquo to

support the mediators (Phillips 2013) ldquoWe have to look for anything thatrsquos an alternative to

violencehellipOn the part of the government we are open to any process that can lower violencerdquo

(Associated Press 2013) and that he had given ldquohis blessing to Emillanirsquos efforts to broker peace

between the gangshelliprdquo (The Daily Herald 2013) However in November 2013 after a general

election the new president Juan Orlando Hernaacutendez through his recently appointed Vice

Minister of Security declared that the government would no longer support the truce process

with the gangs (El Heraldo 2014) Since then the Government of Honduras has not mentioned

the peace process that was initiated in May 2013

Truce making process

It is important to note that prior to the announcement of the gang truce a number of key

stakeholders were somewhat skeptical about its possibility On the one hand some suggested

that a gang truce had been attempted in the past with no success For example one

stakeholder commented to an international media outlet that ldquoEveryone here agrees itrsquos a

positive step forward but people are cautiously optimistic because in 2005 these two gangs

had another peace treaty with each other Now that treaty was very tentative it only lasted

less than two monthsrdquo (Al Jazeera 28 May 2013) On the other hand as noted above other

stakeholders believed that the local gangs did not have enough organizational leadership to

change the behavior of gang members to reduce violence (Bosworth 2013) They argued that

even if gang leaders wanted a gang truce there was no way of enforcing it on the streets

74

Several months prior to the announcement of the gang truce Carlos Mojica Lechuga an 18th

Street Salvadorian gang leader publically stated that representatives of MS13 and 18th Street

in Honduras spoke with several gang truce key stakeholders in El Salvador for the purpose of

replicating the truce in Honduras Tellingly in reflection of the visit Mojica noted that the

advantage of a gang truce is that it formally recognizes Honduran MS13 and 18th Street leaders

as important political persons within the nation He also noted that Honduran gang leaders

have historically been treated poorly and that a gang truce holds the potential for

demonstrating the political power of each of the Honduran gangs (Villiers Negroponte 2013)

The negotiators used different language to describe the early days of the truce Specifically

they mentioned that there had been a consultative process with the gang leaderships so they

were in a process like lighting23 Prior to the truce media sources mentioned that the gang

leaders were offering to stop violence and to not recruit more youth into the gangs24 The gang

leaders also spoke about the ldquopersecutionrdquo they and their family members had suffered during

the previous years highlighting that they had been prohibited social opportunities offered to

others in society (Arce 2013)

Leading up to the negotiations the leadership of both gangs expressed their interest in three

goals 1) lowering violence and crime 2) reconciliation with God society and the government

and 3) helping to improve the social conditions of their communities Although there was not

written documentation on the exact agreement between the parties one MS-13 leader

affirmed that the pact would include all violence (El Comercio 2013) However when talking

about sensitive topics such as extortions which is one of the main sources of income for the

gangs the gang leader said that ldquowould be taken up at a later daterdquo (ABC Internacional 2013)

Leaders of 18th street made similar general statements about ending violence but they were

more specific about their demands One of the 18th Street leaders stated that ldquowhat we want is

23 El Universal 2013 - httpwwweluniversalcominternacional130531obispo-hondureno-descarta-tregua-entre-las-pandillas 24 (Garcia 2013 -httpwwwlaprensahncspmediapoolsitesLaPrensaHondurasSanPedroSulastorycspcid=338546ampsid=276ampfid=98-)

75

to have a dialog with any commission appointed by President Porfirio Lobo and we are sure

that the situation in Honduras will begin to changerdquo (El Nuevo Siglo 2013)

Establishing the gang truce

On May 28th 2013 with public declarations from leaders of both gangs the gang truce was

announced From the beginning of the process the role of the government in the truce was

unclear (La Prensa 2013) Likewise there was little discussion about the exact nature of the

agreement the terms in which gang members would abide and any benefits that would be

made available to those who participated in the truce For example as one stakeholder

indicated the government never decisively considered viable proposals to give the members of

the gangs any opportunities It is important to note that none of the parties signed a formal

commitment and neither MS-13 nor 18th Street signed any type of ceasefire agreement The

gang leaders were in separate locations and were never in direct contact with each other during

the announcement That is both gangs seemingly agreed to the gang truce without ever

talking to each other The ldquopeace processrdquo was publicly announced on May 28 2013 through

ldquojoint but separaterdquo declarations made by the leaders from both gangs imprisoned in the San

Pedro Sula prison (National Penitentiary SPS) (The Daily Herald 2013) The national and

international media widely covered the declarations

The MS-13 leaders said they would not commit any more homicides or any other types of

crimes They ensured that this was an ldquoimmediaterdquo order and would be effective throughout

the country They emphasized ldquoall of the boys know what they have to do starting todayrdquo

(Pachico 2013) The leadership of 18th Street declared that they would stop violence and other

criminal activities but also indicated that the government would have to ldquolisten to themrdquo

Little research has examined whether the gang truce in Honduras ever impacted violence in the

nation Instead anecdotes have been used to portray its effectiveness One facilitator for

example indicated ldquoin Honduras the dialog with the gangs has been positive however the sad

thing in Honduras is that the two main gangs have not accepted a truce between them they

just haven`t accepted it as yetrdquo (La Prensa 2013) despite the fact that ndash in their own words ndash

76

ldquothey do want to hold a dialog with society with the government and with the policerdquo

Conversely gang leaders declared that the truce had been effective For example a leader of

18th Street noted that ldquohellipIt has already done its part telling members in the areas the gang

controls to stop the violence and crimehellip[estimating] crime had already dropped 80 percent in

those areasrdquo (Associated Press June 17 2013) Similarly a member of MS13 estimated that

violence in MS13 controlled areas declined by 45 percent (Associated Press June 17 2013) As

a symbolic gesture of the impact of the truce MS13 leaders also noted that as a gesture of good

will they made and delivered 60 beds for a nursery home in San Pedro Sula (Associated Press

June 17 2013)

Methods

For the present case study we used a pre-test post-test longitudinal quasi experimental design

Data from the 2001 Honduran Census was obtained from the National Institute of Statistics

(INE) These data provided municipal level measures of number of residents population density

per kilometer percent of population who moved in from another municipality ethnicity

percent urban number of residents immigrated to the United States percent female headed

households percent unemployed age composition income percent of households rented and

education level In addition population projections for the years 2005 2010 and 2014 were

also obtained from the National Institute of Statistics (INE) The population levels for the total

population as well as the percentage of residents in a municipality that rural and percentage of

residents who are female were linearly interpolated for the intervening years Examination of

the observed levels indicated that growth was linear overall and so we feel confident that our

linear interpolations are good approximations Second we used municipal level homicide data

by month and year for the period May 2012 through July 2014 These data were provided by

the Honduran National Police through the Honduran US Embassy Both datasets were

merged for the present analysis

Measures

77

The dependent variable for the Honduran case study is the monthly homicide rate which was

calculated by dividing the number of homicides in each municipality by its population and

multiplying this figure by 100000 We examined change by comparing the homicide data 13

months prior to the gang truce to the homicide data 14 months following the truce in each of

the nationrsquos 298 municipalities As presented in Exhibit 21 there were a total of 7910 homicides

over the study period with each municipality averaging 183 homicides (sd=837) A trend

analysis showing the monthly number of homicides on a national level prior to and following

the gang truce is presented in Exhibit 22 It shows that nationally the homicide rate gradually

declined over the study period

Exhibit 21 Summary Statistics

Pre-Truce (n=3809) Post-Truce (n=4101) Total (n=7910)

Mean SD Mean SD Mean SD

Homicide 200 900 166 773 183 837

Population 2874492 8674659 2939557 8870267 2908226 8776124

Homicide rate 521 953 413 814 465 885

Density 9926 10878 9928 10878 9927 10877

Born in the same municipality 8211 1286 8211 1286 8211 1286

Other population group 8545 2390 8545 239 8545 2390

Percent rural 7926 2732 7899 2762 7912 2748

Living in another country 331 650 331 65 331 650

Socioeconomic status 005 097 -002 102 001 100

Exhibit 22 Homicide rate in Honduras by month

78

We used census data to control for several community-level structural factors Principal

components analysis was used to reduce some of the data into a summary measure Exhibit 23

presents the results of the component loadings One component was extracted that we labeled

socio-economic status (SES) This component exhibited high loadings for percent female

headed household percent unemployment and percent completing primary education

Population was used as our exposure variable and population density (per 1 km) residential

stability percentage of the population that is rural percentage of the population that is

indigenous (verify) and the number of residents immigrated to the United States served as our

control variables Population density and outmigration were logged to address skewness in

these two measures

Exhibit 23 Factor Analysis for Socioeconomic Status

Eigenvalue 236093

Variables Factor Loadings

Female-Headed Households 07977

Unemployed 09269

Primary Education 09303

Findings

The results of our t-test analysis are shown in Exhibit 24 It shows that there was a significant

decline in the homicide rate at the municipal level before and after the gang truce The

homicide rate prior to the truce was 697 per 100000 population and following the truce the

homicide rate was 566 homicides per 100000

79

Exhibit 24 T-test on National Data by Month (n= 27 Months)

Homicide Rate Pre Truce Post Truce Change

Mean 6972346 5663592 -1308754

SE 01677361 01151668 02009224

plt 001

Next we examined the effect of the truce through ARIMA regressions The first model in Exhibit

25 does not include an effect of the temporal trend In this model we again find that homicides

declined significantly in the period following the truce We then employed the ARIMA model

again but this time included a variable (month) to control for the temporal trends in the data

Model 2 in Exhibit 25 shows that when we controlled for temporal trends the impact of the

truce we observed was no longer significant Our findings suggest that homicide did not decline

as a consequence of the gang truce but instead the decline in homicides was part of a longer

term historical trend that was independent of the truce

Exhibit 25 Results from the ARIMA Models (n= 27 Months)

Homicide Rate Model 1 Model 2

Effect SE Sig Effect SE Sig

Truce -1309 0350 -0906 0595

m2 -0623 0481

m3 -0153 1447

m4 -0074 0607

m5 -0104 0450

m6 -0050 0466

m7 -0002 0516

m8 0003 0404

m9 0024 1633

m10 -0396 0522

m11 -0754 2424

m12 0837 0366

Time -0030 0047

Intercept 6973 0254 7258 0472

Autocorrelation Coefficient 0010 0408 0111 0249

plt 001

80

Last we used a fixed effects multi-level negative binomial regression to predict the homicide

rate with population as an exposure variable and controlled for the socio-demographic

characteristics of each community The results are presented in Exhibit 26 The only significant

variable in the analysis was the impact of municipal level population stability Specifically we

found that homicide rates increased in areas where residential mobility was high Once again

our analysis showed that time had a negative effect indicating that homicides were decreasing

in all areas over the study period The gang truce itself was unrelated to the decline in

homicides

Exhibit 26 Results from the Multilevel Negative Binomial Models (n= 27 Months)

Homicide Rate Model 1 Model 2

Effect (SE) Sig Effect (SE) Sig

Fixed Effects

Homicide Rate 0004 0004

(0002)

Truce -0119 -0129

(0073) (0072)

Time -0008 -0007

(0005) (0005)

Month 1 0024 0024

(0059) (0059)

Month 2 -0107 -0107

(0062) (0062)

Month 3 -0041 -0041

(0063) (0063)

Month 4 -0033 -0033

(0065) (0065)

Month 5 -0002 -0002

(0067) (0067)

Month 6 (reference)

Month 7 0027 0027

(0051) (0051)

Month 8 0023 0023

(0057) (0057)

Month 9 0036 0036

81

(0056) (0056)

Month 10 -0051 -0051

(0057) (0057)

Month 11 -0031 -0031

(0058) (0058)

Month 12 0187 0187

(0056) (0056)

Born in the same municipality -0013

(0003)

Pecent rural -0003

(0002)

Socioeconomic status 0029

(0044)

Percent dominant population 0001

(0002)

Density (ln) -0010

(0050)

Living in another country (ln) 0001

(0061)

Intercept -10042 -10051

(0061) (0058)

ln(Alpha) -2368 -2367

(0148) (0148)

Random Effects

Var(Truce Effect) 0066 0066

(0024) (0025)

Var(Intercept) 0388 0309

(0049) (0042)

Cov(truce effect intercept) -0024 -0006

(0027) (0025)

plt 001

82

Conclusions

Violent crime in Honduras is widespread With a homicide rate of about 86 per 100000

population Honduras is one of the most violent nations in the world (United Nations 2013)

This compared to an average homicide rate around the world of 62 per 100000 people and

about 27 per 100000 in Central America (United Nations 2013) Given the high rate of violence

in Honduras it is clear that new and innovative ways of reducing violence should be considered

This case study gave an overview of the implementation and impact of the gang truce

experience in Honduras in 2013

The goal of the truce was to significantly reduce the number of homicides In order to assess

the impact of the gang truce in Honduras a pre-test post-test longitudinal quasi-experimental

design was used Census and homicide data were merged at the municipal level to asses

whether the truce had an impact on homicides controlling for population characteristics and

the natural trend in violent crime

Overall the findings suggest that while the homicide rate in Honduras was on a slight

downward trend the gang truce itself was unrelated to any homicide reductions That is the

gang truce had no measurable impact on homicides in Honduras Given the drastic reductions

achieved in El Salvador and the fact that Hondurasrsquo gang truce was a replication of El

Salvadorrsquos the following discussion will examine two important differences between the two

countries gang truces First the implementation of the truce in Honduras did not appear to

obtain trust between all parties involved and did not achieve any notable short-term

deliverables That is the implementation was not robust Second some suggest that the gangs

in Honduras do not have the organizational sophistication to be able to control their members

on the streets rendering them incapable of carrying out any truce agreements

First the implementation of the truce in Honduras did not appear robust The communication

between the parties was weak and none of the parties completed any significant actions as part

of the truce The Catholic Church and OAS served as the bridge between the two main gangs in

Honduras MS13 and 18th Street and the executive branch of the government The ldquopeace

83

processrdquo as it was called in Honduras was initiated in May 2013 After the general election in

November 2013 the newly appointed administration declared that it would no longer support

the truce process with the gangs As a result the truce was short-lived and the governmental

support for the effort shifted with the change in leadership

Though the negotiations were largely based on the good will of a well-respected negotiator

few tangible incentives were offered during the process The negotiations from the beginning

involved discussions about large scale social programming Given the short time frame these

goals in hindsight were unrealistic Naturally there was little trust between the parties at the

beginning of the negotiations and without any quick tangible deliverables from either side the

truce never really materialized

The second challenge to implementing a successful gang truce in Honduras might have been

related to the nature of the gangs themselves It was unclear whether the gangs possessed the

level of cohesion and hierarchical leadership required to make some measures feasible If the

gangs do not have the organizational capacity to control their members on the streets any

agreements that come from the negotiations would be difficult to implement In the earlier

case study of the Salvadorian experience it was suggested that the successes in that country

were related to one of the gangrsquos organizational capacity to impose control of its members The

gangs in Honduras might be different There is at least some evidence that MS 13 in Honduras

might have less organizational leadership (Bossworth 2013) less control over turf (Latin News

Daily Report 2013) and have been more organizationally fractured (Dudley 2013) than their

counterparts in El Salvador The gangrsquos ability to operate as an efficient organization can greatly

impact the outcome of the truce process In general we know that gangs do not have very high

levels of organizational sophistication (Decker Katz and Webb 2008 Decker Bynum Weisel

1998) It might be that gangs in Honduras are more the norm in terms of organizational

capacity

In summary the 2012 truce negotiations in Honduras did not produce any measurable

reductions in the homicide rate The gangs wanted to speak to Honduran society and they even

84

preferred to speak with the governmental authorities but they never interacted with either

The negotiation process seemed to end as quickly as it started The gangs did not deliver with

lower rates of violence and the government did not provide social programs It would be safe to

say that a robust gang truce did not materialize in Honduras The implementation of the truce

seemed to struggle for two primary reasons First the mediators were not able to accomplish

any quick wins to build trust between the parties involved The commitment level on all sides

was not clear throughout the process Second it is not clear whether the gangs in Honduras

have the organizational capacity to control members on the street as would be required to

carry out an effective gang truce In short the 2013 gang truce in Honduras was unsuccessful

85

Conclusions Policy Implications and Recommendations

The purpose of this report was to systematically examine the effectiveness of gang truces Gang

truces have been widely implemented but rarely evaluated Of those gang truces that have

been evaluated little attention has been given to why and how they came into existence In

this report we reviewed prior research on gang truces and presented case studies of gang

truces implemented in El Salvador Jamaica and Honduras for the purpose of understanding

the negotiation processes undertaken with and between gangs and other stakeholders We

were interested in identifying the actors involved in the negotiations the goal(s) of the

negotiations and the strategies employed to carry them out Most importantly however we

wanted to determine whether the gang truce resulted in a reduction in the number of

homicides Each case study offers lessons learned that are unique to their particular

circumstances and when considered together provide direction to policymakers on the

benefits and risks of implementing gang truces

The case studies presented here constitute the most comprehensive evaluations of gang truces

to date Existing documents were used to collect information about the processes associated

with each gang truce Many of these documents included such items as peer reviewed articles

books and reports The majority of these documents were collected over the Internet

requesting documents from those close to the truce and searching library databases Related

the case studies made use of articles obtained from local newspapers The newspaper articles

were not only intended to provide a historical record of the development of each gang truce

but also to provide additional insight into the various external forces that might have impacted

the gang negotiations Because the newspaper serves as a forum for the community to speak

about its concerns newspaper articles also provided a rich source of data on how those in the

community felt about the gang truce Accordingly the newspaper articles offered a different

view of the problem and offered different opinions as to how a gang truce should or should not

be implemented We also conducted a small number of in-depth qualitative interviews with

key informants These data were collected to supplement existing documents and to clarify

issues associated with the negotiation processes This included but was not limited to

86

questions pertaining to identifying the actors involved in the negotiations the goal(s) of the

negotiations and strategies employed to carry out negotiations The interviews were intended

to obtain information from those who possessed first hand knowledge about the gang truce in

each nation

We examined the impact of each truce using official data We first performed a simple t-test

comparing the homicide rates before and after the truce However as discussed above this

technique has limitations The most severe of which is that even if the test was significant it

would be difficult to determine whether the difference was due to the gang truce or a natural

change over time in the outcome We addressed this limitation by using time series models

whereby the homicide rate for the community was modeled as a function of time with truce

period indicators included to measure the effect of the truce net of the temporal trends These

models were estimated with ARIMA techniques Supplemental models were also employed to

examine and control for factors other than the truce that might have impacted homicide over

the study period

Summary of Findings Related to the Implementation of a Gang Truce in the Three Sites

We found that the implementation of gang truces have a number of common characteristics

The first is that in each case a community was experiencing an uncharacteristically high number

of gang related homicides over a fairly lengthy period of time The continued high level of

violence in each case resulted in the community placing strong pressure on the government in

general and the justice system in particular to respond to the problem quickly and effectively

In each case they had first attempted to control gang violence through suppression oriented

strategies and these strategies were found to be ineffective over the intermediate and long

term In turn each communityrsquos inability to exercise traditional informal and formal social

control to decrease levels of violence became self evident to the public and government This

resulted in both the state and community to seek (or participate in) an alternative strategy in

which negotiators would formally andor informally work with gang leaders to establish a truce

that would reduce gang homicide

87

Key stakeholders involved in the negotiation and establishment of each gang truce were fairly

similar In each of the cases examined the gang leaders of the largest and most violently

involved gangs were willing to consider participating in negotiations that could lead to a truce

In each of these cases it was clear that the gangs not only sought to collaborate with the

negotiators for the purpose of reducing violence but perhaps more importantly were seeking a

means in which to gain greater more positive recognition in the community and to reap some

form of benefit to themselves their members and possibly their community In each case

while not always formally involved government officials were at a minimum made aware of

negotiations and in some cases solicited the assistance of third partyrsquos to broker an agreement

between stakeholders In each case it was at least implicitly understood that the government

would ldquolistenrdquo to the gang leaderrsquos expectations of the government and what theymdashthe gang

leaders--had to offer in exchange We found that when the government was no longer willing

to ldquolistenrdquo to or collaborate with negotiators the truce processes ended abruptly Negotiators

were typically comprised of a very small group (ie 2-3) of individuals who were perceived to

be ldquohonest brokersrdquo In El Salvador and Honduras this included a high ranking Catholic Church

official a leader from an international diplomatic organization (ie OAS) and other neutral

parties In Jamaica this included a quasi-governmental organization that had been established

for the purpose of brokering negotiations between gangs to reduce violence and the local

university which had access to staff who were perceived to be neutral but had an interest in

reducing violence due to its proximity to the university

The strategies used to execute each gang truce were similar but yet importantly different They

were similar in that each involved a team of negotiators working to identify common goals to

be achieved and identifying tangibles that could be delivered to the gang leaders gang

members and their community in exchange for the gang achieving their stated goals They

were different however in terms of the structure of the delivery of each parties promise to the

other In Honduras and Jamaica it appears that gang leaders committed to reducing gang

violence in exchange for general promises made by the negotiators for example that

substantial public works programs would be implemented for the goal of reducing

unemployment among gang members and the community In both of these cases it required

88

the government to develop and deploy large scale social programming in a very quick period of

timemdashsomething that neither government had a strong record of demonstrating In El Salvador

negotiators employed a strategy of the gang leaders promising to deliver immediate changes in

gang member behavior for immediate administratively natured changes by the government

For example in exchange for a reduction in gang violence the government agreed to relocate

imprisoned gang leaders to less restrictive prisons and provide them some privileges Following

the successful execution of the first part of the truce which resulted in near term success for

both parties they began to negotiate issues that would take a longer period of time for the

gangs and government to deliver Our findings suggested that some promised deliverables need

to be easily and quickly delivered early in the process so that trust increases between both

parties Stakeholders only have a brief period of time to provide promised benefits before trust

is lost and that tangible benefits need to be delivered in weeks or months not years

Summary of Findings Related to the Impact of Gang Truces in the Three Sites

Our findings suggest that El Salvadorrsquos gang truce had a significant and dramatic impact on the

homicide rate The mean number of monthly homicides declined from about 354 prior to the

truce to about 218 following the truce for a net decrease of about 136 homicides per month

Our forecast showed that between March 2012 and June 2014 the truce had saved about 5501

lives However there was no significant change between the pre-truce and post-truce periods

in the number of thefts extortions robberies rapes and auto theftsrobberies We also found

that the gang truce did not result in a homogenous decline in violence across municipalities

About 61 percent of municipalities experienced a decline in homicides but the decline in

violence varied substantially between municipalities We examined this issue further by parsing

out the relative influence of the number of MS13 and 18th Street gang members on the street

and in prison from each municipality Our analyses indicated that following the truce the

number of MS13 and 18th Street gang members on the street in a municipality was not

significantly related to a decline in homicide but the number of imprisoned MS13 and 18th

Street gang members was associated with a significant change in homicides following the gang

truce In particular the number of imprisoned MS13 gang members was associated with a

89

significant decline in homicides following the gang truce and the number of imprisoned 18th

Street members was associated with a significant increase in homicides following the truce

In Jamaica our initial findings showed that the gang truce might be an effective mechanism for

reducing violence Bivariate analyses showed a significant decline in homicides immediately

after the truce was implemented This explains the work previously published by policymakers

researchers and news reporters Upon further examination of the data however comparing

change in the target and comparison areas and accounting for temporal trends we found that

the decline in homicide was part of a larger nationwide decline in violence and that the gang

truce was not responsible for the decline The only significant effect that we uncovered was the

possibility that homicides were displaced outside the target area for a brief period of time but

then returned to normal

Our findings from Honduras told a similar story as Jamaica Initial analysis showed that the

number of homicides on average declined across municipalities following the gang truce

Specifically the mean number of homicides declined by 13 per 100000 population with 687

homicides per 100000 population on average occurring in each municipality prior to the truce

and 566 homicides per 100000 population on average occurring in each municipality after

the truce However after we examined the effect of the truce through the ARIMA model and

included a variable (month) to control for the temporal trends in the data the impact of the

truce we observed in our bivariate analysis was no longer significant Our findings as in

Jamaica suggested that the decline in homicides was not the consequence of the gang truce

but instead the decline in homicides was part of a long term decline in homicides due to

exogenous factors

The Potential Benefits and Consequences of a Gang Truce

Over the last several years there have been a number of naturally occurring experiments

involving gang truces in a variety of nations in various regions of the world Findings from

evaluations of gang truces are mixed As noted above in El Salvador the gang truce could be

characterized as highly effective at least for the two years following the truce It is worth

mentioning that even after the truce breakup homicides rates while above truce levels

90

continue slightly below pre-truce levels In Jamaica and Honduras the gang truce had no impact

on violence In Los Angeles and Trinidad there was evidence that violence decreased for at

least ninety days but then increased substantially beyond those rates observed prior to the

gang truce (see the introduction section of this report for this discussion) As a consequence it

appears that the potential for long term consequences might out weigh the potential for short

term benefits Only one study site(El Salvador) demonstrated a truce having a substantial and

long term impact on violence Others conversely demonstrated the truce had no impact or

increased violence over the long term In fact a number of scholars have noted that gang truces

are likely to result in a boomerang effect with gang violence increasing over the long run

because of enhanced cohesion within the gang (Klein 1995) Maguire (2013) notes that when

government officials negotiate a truce with gangs they might ldquoinadvertently be acknowledging

gangs as legitimate social entitiesrdquo (p 11) This in itself might increase cohesion among gangs

which has been found to be associated with increased levels of criminality (Decker et al 2008

Klein 1971 Maguire 2013)

Further research is needed to examine how gang truces might impact group cohesion and if it

does whether this in turn results in greater violence Gang truces convey the well-intentioned

image that violence has been addressed and policymakers are doing something about the

problem but researchers need to better understand the probability of a gang truce reducing

violence increasing violence or having no impact This will better position policymakers to

understand the relative risks associated with these types of interventions

Our findings also suggest that while gang truces could be an effective intervention in areas

where gangs are highly structured and organized such as El Salvador they could be counter-

productive in areas where gangs are not as structured and organized Because the vast majority

of street gangs are not well organized (Klein 1995 Spergel 1995) the utility of a gang truce in

reducing violence might be limited Our findings coupled with prior research suggest that gang

interventions need to be tailored to the nature of the gang and its members or it risks

increasing gang violence

Final Thoughts

91

Our analysis suggests that gang truces should only be used as a means of last resort and then

only under certain conditions Given the risks associated with a gang truce communities with

high levels or at least modest levels of formal social control should rely on other more

promising gang control strategies Only when the state has limited or greatly reduced capacity

for social control should a truce be considered Concomitantly a gang truce should only be

considered when a community is experiencing a substantial amounts of gang violence

Communities that are experiencing minimal to modest amount of gang violence may have more

to lose from the establishment of a gang truce than they have to gain Additionally our

findings suggest that a gang truce might only be feasible when gangs are sufficiently organized

to the extent that they have the capacity to regulate memberrsquos behavior In other words gang

leaders must have the ability to reduce their memberrsquos involvement in violence for a gang truce

to work Our findings more concisely suggest that gang truces should only be considered when

there are a great number of gang homicides the state has limited capacity to address the

problem and gang leaders have enough informal social control over their members that they

themselves can substantially control the levels of violence in their community by regulating

their memberrsquos behavior

Recommendations

Gang truces are conjunctural strategies States who suffer from gang-related violence must

establish permanent public policies for crime control and prevention A government that

considers implementing a gang truce should be aware that it cannot become the center

strategy of its public policy for citizen safety

Gang truces should only be used as a means of last resort and then only under certain

conditions Stakeholders must determine whether a process of dialog or negotiation with gangs

is legal ethical and feasible

Stakeholders must anticipate demands that are likely to arise and their response options Some

demands may be easily met such as improved prison conditions Others are much more

difficult and amorphous such as community development through more integrated violence

92

prevention programs (such as those implemented by SolucionES in El Salvador) local economic

development programs or economic reinsertion of ex-gang members

Stakeholders should incorporate immediately achievable and demonstrable deliverables Long-

term goals and promises are unlikely to create the trust needed to sustain a gang truce

Stakeholders must first determine the position in which they are negotiating the incentives that

are possible to deliver and the boundaries and limits they face Gangs are mostly likely to trust

representatives from NGOs community-based organizations and members of the faith based

community as brokers because they are considered more reliably neutral advocates for peace

They need to understand the capacity of the government to deliver promises in a timely

manner

Governments have to make a choice about the visibility and transparency of its participation

This decision needs to be made in the context of the national and local laws the publicrsquos

expectations of transparency and patterns of practices of the past

Governments must be strategic in their support for truce initiatives Some donor funded

programs run by the government prohibit gang member participation and if the government

does not receive approval from the donor it may risk the donor withdrawing its sponsorship of

the program

Governments must ensure an inter-institutional coordination for the management of truces to

avoid the responsibility to be of a single government institution It is necessary to generate or

collect reliable and pertinent data that can be used to analyze and assess the process

It is necessary to implement an effective monitoring system of the truce process similar to that

used in the full report as it can help parties understand what is working and who is delivering

on their promises More specifically the monitoring and truce management system should be

able to identify truce violations and be prepared to respond through the use of legal and

effective practices if stakeholders do not comply

93

Finally it is necessary to develop evaluations of gang truces and monitoring programs and

support violence prevention activities local economic development activities and pilot

programs to support the reinsertion of ex-gang members into society Clearly national

governments municipal governments NGOs and community-based organizations need

increased capacity and resources to discourage the growth of gangs among at-risk youth It is

therefore increasingly important to create economic opportunities for gang members willing to

leave the gangs and find other legal employment Developing and sustaining those

opportunities in nations with high incidences of poverty will require significant international

funding

94

References

Anderson Elijah 1999 Code of the Street New York Norton

mdash 1998 ldquoThe Social Ecology of Youth Violencerdquo In Crime and Justice vol 24 edited by M

Tonry 65-104

mdash 1990 Street Wise Chicago University of Chicago Press

Arce A (2013 May 28) Honduras gangs declare truce ask talks with govt Associated Press

The Big Story Retrieved from

httpbigstoryaporgarticlehonduras-gangs-declare-truce-seek-talk-govt

Archibold Randal 2012 ldquoGangsrsquo Truce Buys El Salvador a Tenuous Peacerdquo The New York Times

Originally published online on August 27 2012

Axelrod R (1986) ldquoAn Evolutionary Approach to Normsrdquo American political science review

80(04) 1095-1111

Ayala Edgard 2012 Gangs Back Plan for Violence-Free Districts in El Salvador

httpwwwipsnewsnet201212gangs-back-plan-for-violence-free-districts-in-el-

salvador Found on November 4 2014

Bakrania Shivit 2013 Policy Responses to Criminal Violence in Latin America and the

Caribbean GSDRC Helpdesk Research Report 934 Birmingham UK GSDRC University

of Birmingham

Bargent James 2013 ldquoRise in Disappearances Feeds Doubts Over El Salvador Trucerdquo August 1

2013 Found at httpwwwinsightcrimeorgnews-briefsrise-in-disappearances-in-el-

salvador-feeds-doubts-over-truce on September 1 2014

Battin Sara R Karl G Hill Robert D Abbot Richard F Catalano and J David Hawkins 1998

ldquoThe Contribution of Gang Membership to Delinquency Beyond Delinquent Friendsrdquo

Criminology 36 93-115

Block Richard 2000 ldquoGang Activity and Overall Levels Of Crimerdquo Journal of Quantitative

Criminology 16 (3) 369-83

iii

or broader trends related to violence reductions 3) a failure to examine the processes involved

in creating gang truces 4) a failure to examine unintended consequences of the truces beyond

their impact on gun violence and 5) a lack of sophisticated statistical analysis capable of

discerning overall trends cyclic patterns outliers and turning points

Overview of Study Design

This policy brief examines gang truces implemented in El Salvador Jamaica and Honduras

including the processes undertaken with and between gangs and other stakeholders We

collected information about the processes associated with each gang truce from a variety of

sources including peer reviewed articles books reports local newspaper articles and

interviews of persons with first-hand knowledge about the gang truce in each nation We

examined the impact of each truce using official data We first performed a simple t-test

comparing the homicide rates before and after the truce We then employed time series

models not used in other studies to evaluate homicide rates as a function of time with truce

period indicators included to measure the effect of the truce net of the temporal trends We

also employed supplemental models to examine and control for factors other than the truce

that might have affected homicides over the study period

Research Findings

Implementing a Gang Truce

The gang truces studied have a number of common characteristics First in each case a

community was experiencing an uncharacteristically high number of gang related homicides

over a fairly lengthy period of time which resulted in each community placing strong pressure

on the government in general and the justice system in particular to respond to the problem

quickly and effectively Second each community had first attempted unsuccessfully to control

gang violence through suppression-oriented strategies Third each communityrsquos inability to

exercise traditional informal and formal social control to decrease levels of violence became

self-evident to the public and government This resulted in all involved stakeholders wanting to

seek (or participate in) an alternative strategy in which brokers would formally andor

informally work directly with gang leaders to establish a truce that would reduce homicides

iv

Key stakeholders involved in the negotiation and establishment of each gang truce were also

somewhat similar In each case examined the leaders of the largest and most violently

involved gangs were willing to participate In each of these cases it was clear that the gangs

sought to collaborate with the brokers not only for the purpose of reducing violence but

perhaps more importantly as a means to gain greater more positive recognition in the

community and to reap some form of benefit to themselves their members and perhaps their

community In each case while not always formally involved government officials were at a

minimum made aware of negotiations and in some cases solicited the assistance of third

parties to broker an agreement between stakeholders In each case it was at least implicitly

understood that the government would ldquolistenrdquo to the gang leaderrsquos expectations and offers

We found that when the government was no longer willing to ldquolistenrdquo to or collaborate with

brokers the truce processes ended abruptly Brokers were typically comprised of a very small

group (ie 2-3) of individuals who were perceived to be ldquohonest brokersrdquo In El Salvador and

Honduras this included a high ranking Catholic Church official a leader from an international

diplomatic organization (ie OAS) and other neutral parties In Jamaica this included a quasi-

governmental organization that had been established for the purpose of brokering negotiations

between gangs for the purpose of reducing violence as well as the local university which had

access to staff who were perceived to be neutral but had an interest in reducing violence near

the university

The strategies used to execute each gang truce were generally similar but had important

differences Each involved a team of brokers working to identify common goals to be achieved

and tangibles that could be delivered to the gang in exchange for the gang achieving the stated

goals They differed by how each partyrsquos promise was delivered In Honduras and Jamaica gang

leaders traded violence reduction pledges for long-term government and social changes such

as the development of substantial public works programs to reduce unemployment In

Honduras and Jamaica the government was asked to develop and deploy large scale social

programming in a short time framemdashsomething for which neither country had a strong track

record In El Salvador by contrast brokers secured promises for immediate changes in gang

behavior in exchange for feasible immediate deliverables from the government For example in

v

exchange for a reduction in gang violence the government agreed to immediately relocate

imprisoned gang leaders to less restrictive prisons and provide them some privileges Following

the successful execution of the first part of the Salvadoran truce which resulted in near term

success for both parties they began to negotiate broader issues that would take longer for both

sides to deliver Our findings suggest that some promised deliverables need to be easily and

quickly delivered by both parties early in the process to achieve trust and serve as a first test

of gang leadersrsquo ability to deliver Stakeholders have only a brief period of time to provide

promised benefits before trust is lost meaning that tangible benefits need to be delivered in

weeks or months not years

The Impact of a Gang Truce

El Salvador We found that El

Salvadorrsquos gang truce had a definite

impact on the homicide rate The

mean number of monthly homicides

declined from about 354 prior to the

truce to about 218 following the

truce for a net decrease of about

136 homicides per month Our data

show that between March 2012 and

June 2014 the truce had saved

about 5501 lives (see exhibit 1)

From a hypothetical stand it is

possible to make the assumption that a number of these deaths averted could have been

transformed in disappearances and therefore they were not counted within the official

homicide statistics However the results from the analysis point out that the number of

disappearances was not significantly related with the change in the global behavior of

homicides Additionally over the period of analysis (January 2010 to June 2014) there was no

significant change between the pre-truce and post-truce periods in the number of thefts

extortions robberies rapes and auto theftsrobberies

Exhibit 1 Forecast of homicides without gang truce

vi

We also found that the gang truce did not result in a homogenous decline in violence across

municipalities About 61 percent of municipalities experienced a decline in homicides but the

decline in violence varied substantially between municipalities We studied this issue further by

examining the impact of the initiative ldquoFree Violence Municipalitiesrdquo and it was found that the

behavior of violence in those municipalities was not significantly linked to the initiative but

rather to the general dynamic of the truce process nationwide Additionally we parsed out the

relative influence of the number of MS13 and 18th Street gang members on the street and in

prison from each municipality Our analyses indicated that following the truce the number of

MS13 and 18th Street gang members on the street in a municipality was not significantly

related to a decline in homicide but the number of imprisoned MS13 and 18th Street gang

members from the municipality was In particular the number of imprisoned MS13 gang

members from a municipality was associated with a significant decline in homicides in that

municipality following the gang truce and the number of imprisoned 18th Street members

from a municipality was associated with a significant increase in homicides in that

municipality following the truce These findings lend support to the idea that MS13 is more

organized than the typical street gang and that imprisoned MS13 gang members exhibit strong

influence over their fellow gang members on the street Our findings also suggest however

that the gang truce had a boomerang effect in municipalities with high numbers of imprisoned

18th Street members implying that 18th Street might not have as much organizational capacity

to regulate violence on the streets as MS13 The truce provided incarcerated MS13 and 18th

Street gang leaders an opportunity to negotiate with high-ranking officials and influential

diplomats including representatives of the Organization of American States This may have

increased their legitimacy inside and outside of their gangs It appears that MS13 was able to

exert its span of control over the communities in which they had influence and they were able

to deliver on the terms of the gang truce negotiations In the case of 18th Street however

incarcerated gang members may not have had the same organizational capacity for

communicating and carrying out directives In fact a review of the gang truce indicated that

there was a conflict taking place between two factions within 18th Street Consequently the

organizational structure and culture of 18th Street might be more diffuse than that of MS13

vii

and its leadership structure might not be as strong because of the internal fractures within the

gang This might further explain why homicides increased in 18th Street communities The

internal fractures within the 18th Street gang may have resulted in intra-gang violent conflict

that was largely contained within 18th Street controlled territories

Jamaica In Jamaica at first glance our impact findings appeared to show that the gang truce

might be an effective mechanism for reducing violence Bivariate analyses showed a significant

decline in homicides immediately after the truce was implemented This explains the work

previously published by policymakers researchers and news reporters Upon closer

examination of the data however comparing change in the target area to other areas in

Jamaica and accounting for temporal trends we found that the decline in homicide was part of

a larger nationwide decline in violence and that the gang truce was not responsible for the

decline The only significant effect that we uncovered was the possibility that homicides were

displaced outside the target area for a brief period of time but then returned to previous

levels

Honduras Our impact findings from our analysis of data from Honduras tell a very similar story

as Jamaica Initial analysis showed that the number of homicides on average declined across

municipalities following the gang truce Specifically the mean number of homicides declined by

12 per 100000 population from an average of 687 per 100000 population in each

municipality before the truce to an average of 566 thereafter However after we examined the

effect of the truce through time series analysis and included a variable (month) to control for

the temporal trends in the data the impact of the truce we observed in our bivariate analysis

was no longer significant Our findings as in Jamaica suggest that the decline in homicides was

less a consequence of the gang truce than of a broader short to medium term trend And

importantly in both Jamaica and Honduras the respective governments failed to deliver on

gangsrsquo demands for large-scale social and employment programs

Conclusions and Policy Implications

Over the last several years there have been a number of naturally occurring experiments

involving gang truces in a variety of nations and regions Findings from evaluations of gang

viii

truces are mixed In El Salvador the gang truce could be characterized as an effective short

term strategy to reducing homicides It is worth mentioning that despite homicides rates are

above truce levels they continue slightly below pre-truce levels In Jamaica and Honduras the

gang truce had no short or medium-term impact on overall violence In Los Angeles and

Trinidad y Tobago there was evidence that violence decreased for at least ninety days but then

increased substantially beyond those rates observed prior to the gang truce Thus the

effectiveness of the truce in El Salvador appears to be isolated and must be evaluated within

the context of other truces that have failed to reduce homicide violence Policy makers must

evaluate whether the conditions that allowed short term effectiveness of the gang truce in El

Salvador (such as the ability to promise and deliver immediate results) exist in other violent

areas before evaluating whether a truce strategy might be appropriate And they should be

heavily cautioned that the potential for long term negative consequences might outweigh the

potential for short term benefits

Indeed it is important to note that a number of scholars have noted that gang truces are likely

to result in a boomerang effect with gang violence increasing over the long run because of

enhanced cohesion within the gang (Klein 1995) Maguire (2013) notes that when government

officials negotiate a truce with gangs they might ldquoinadvertently be acknowledging gangs as

legitimate social entitiesrdquo (p 11) This in itself might increase cohesion among gangs which has

been found to be associated with increased levels of criminality (Decker et al 2008 Klein 1971

Maguire 2013) Hence it is important to consider the fact that gangs are illegal groups in El

Salvador and it should be cautious when carrying out dialog or negotiation processes with

them Further research is needed to examine how gang truces might impact group cohesion

and if it does whether the cohesion created could be effectively directed toward more

productive non-violent endeavors Gang truces convey the well-intentioned image that violence

has been addressed and policymakers are doing something about the problem but unless the

truce is implemented in a manner and under conditions where immediately achievable results

can be promised delivered and measured there remains a significant chance that the truce will

fail or worse yet backfire Thus it is imperative that any type of concession made by

ix

Governments to gangs within a truce framework should be transparent so that all sectors of

society have certainty that every action is being done within the existing rule of law2

In the case of El Salvador the truce arises from to the absence of effective public policies and

practices for violence control and prevention The truce was planned as a strategy to reduce

gang-related homicides During the process different organizations got involved including

religious (facilitating and protecting human rights) non-government (managing and facilitating

dialog and negotiation processes) international (providing funding for insertion programs) and

government (facilitating and providing certain conditions for dialog and negotiation)

organizations Some of the concessions that the Government provided in order to achieve a

reduction of homicides were within the law but others generated confusion and they seemed

to be close to the legal or socially acceptable limits This fact along with the poor transparency

of authorities in the management of the practice with public media and public weakened the

process and postponed its continuity

The present study suggests that gang truces should only be used as a means of last resort

and then only under certain conditions Given the risks associated with a gang truce

communities with high levels or at least modest levels of formal social control should rely on

other more promising gang control strategies such as pulling levers (ie Boston Ceasefire)

community oriented policing and the Gang Resistance Education and Training (aka GREAT)

program Only when the state has limited or greatly reduced capacity for social control should a

truce be considered Concomitantly a gang truce should be considered as an alternative only

when a community is experiencing a substantial amount of gang violence Communities that

are experiencing minimal to modest amount of gang violence may risk more from the

establishment of a gang truce than they have to gain Additionally our findings suggest that a

gang truce might only be feasible when gangs are sufficiently well organized to be able to

regulate their membersrsquo behavior and cause their members to behave less violently In El

22 Pragmatically this issue is complicated On the one hand transparency is a foundational element within a democracy and is necessary to ensure proper oversight of the government On the other hand it might not be possible to implement a gang truce with too many actors having a voice Policymakers might consider creating a policy that allows such negotiations take place but requires particular actors (such as a judicial body) to be informed of the process to ensure transparency and adherence to the rule of law

x

Salvador there is evidence of the strong organizational structure of gang MS13 Among other

factors the magnitude of its membership the chain of command from its leaders in prison and

the discipline of its leaders in the streets seemed confirmed On the contrary the

organizational structure of gang Barrio 18 ndash divided in two factions fighting over the leadership

ndash showed to be a less stable counterpart within the truce

Finally dialog and negotiations processes with or between gangs must have the capacity to

promise and deliver immediate benefits to the gangs that gang members can see or experience

in order to secure their continued participation in the truce as well as the capacity to monitor

and respond to truce violations Most importantly any effort aimed at reducing violence is

important and should be examined and assessed but it must have a transparent foundation

especially when it affects population rights as a whole

General recommendations

Gang truces are conjunctural strategies States who suffer from gang-related violence must

establish permanent public policies for crime control and prevention A government that

considers implementing a gang truce should be aware that it cannot become the center

strategy of its public policy for citizen safety

Gang truces should only be used as a means of last resort and then only under certain

conditions Stakeholders must determine whether a process of dialog or negotiation with gangs

is legal ethical and feasible

Stakeholders must anticipate demands that are likely to arise and their response options Some

demands may be easily met such as improved prison conditions Others are much more

difficult and amorphous such as community development through more integrated violence

prevention programs (such as those implemented by SolucionES in El Salvador) local economic

development programs or economic reinsertion of ex-gang members

Stakeholders should incorporate immediately achievable and demonstrable deliverables Long-

term goals and promises are unlikely to create the trust needed to sustain a gang truce

Stakeholders must first determine the position in which they are negotiating the incentives that

are possible to deliver and the boundaries and limits they face Gangs are mostly likely to trust

xi

representatives from NGOs community-based organizations and members of the faith based

community as brokers because they are considered more reliably neutral advocates for peace

They need to understand the capacity of the government to deliver promises in a timely

manner

Governments have to make a choice about the visibility and transparency of its participation

This decision needs to be made in the context of the national and local laws the publicrsquos

expectations of transparency and patterns of practices of the past

Governments must be strategic in their support for truce initiatives Some donor funded

programs run by the government prohibit gang member participation and if the government

does not receive approval from the donor it may risk the donor withdrawing its sponsorship of

the program

Governments must ensure an inter-institutional coordination for the management of truces to

avoid the responsibility to be of a single government institution It is necessary to generate or

collect reliable and pertinent data that can be used to analyze and assess the process

It is necessary to implement an effective monitoring system of the truce process similar to that

used in the full report as it can help parties understand what is working and who is delivering

on their promises More specifically the monitoring and truce management system should be

able to identify truce violations and be prepared to respond through the use of legal and

effective practices if stakeholders do not comply

Finally it is necessary to develop evaluations of gang truces and monitoring programs and

support violence prevention activities local economic development activities and pilot

programs to support the reinsertion of ex-gang members into society Clearly national

governments municipal governments NGOs and community-based organizations need

increased capacity and resources to discourage the growth of gangs among at-risk youth It is

therefore increasingly important to create economic opportunities for gang members willing to

leave the gangs and find other legal employment Developing and sustaining those

opportunities in nations with high incidences of poverty will require significant international

funding

xii

References

Cotton Paul 1992 ldquoViolence Decreases with Gang Trucerdquo JAMA 268(4) 443-44

Decker Scott H 2003 Policing Gangs and Youth Violence Belmont CA Wadsworth

Decker Scott H and W Reed 2002 Responding to Gangs Evaluation and Research

Washington DC National Institute of Justice

Decker Scott H Charles M Katz and Vincent J Webb 2008 ldquoUnderstanding the Black Box of

Gang Organization Implications for Involvement in Violent Crime Drug Sales and

Violent Victimizationrdquo Crime and Delinquency 54 153-72

Esbensen Finn-Aage and D W Osgood 1997 ldquoNational Evaluation of GREATrdquo US

Department of Justice Office of Justice Programs National Institute of Justice

Katz Charles M and Webb Vincent J 2006 Policing Gangs in America New York Cambridge

University Press

Klein Malcolm W 1995 The American Street Gang New York Oxford University Press

mdash 1971 Street Gangs and Street Workers Englewood Cliffs NJ Prentice-Hall

Maguire Edward Charles Katz and David Wilson 2013 ldquoThe Effects of a Gang Truce on Gang

Violencerdquo Unpublished paper Washington DC American University

Maguire Edward 2013 Research Theory and Speculation on Gang Truces Woodrow Wilson

International Center for Scholars in Washington DC

McCorkle R and T Miethe 2002 Panic The Social Construction of the Street Gang Problem

Upper Saddle River NJ Prentice-Hall

Ordog Gary J W Shoemaker Jonathan Wasserberger and Michael Bishop 1995 ldquoGunshot

Wounds Seen at a County Hospital Before and After a Riot and Gang Truce Part Twordquo

Journal of Trauma-Injury Infection amp Critical Care 38(3) 417-19

Ordog Gary J Jonathan Wasserberger Julius Ibanez Michael Bishop Eduardo Velayos

Subramaniam Balasubramanium and William Shoemaker 1993 ldquoIncidence of Gunshot

Wounds at a County Hospital Following the Los Angeles Riot and a Gang Trucerdquo Journal

of Trauma 34 779-82

Whitehill J M Webster D W amp Vernick J S 2012 ldquoStreet conflict mediation to prevent

youth violence Conflict characteristics and outcomesrdquo Injury prevention injuryprev-

2012

1

Introduction

Given the increasingly devastating individual family and community effects of gang activity

over the past several decades an increasing body of literature has focused on gangs gang

members and gang activity A core theme running throughout this body of literature is that

gang members are significantly more likely to be the offenders and victims of violent crime than

non-gang members (Battin Hill Abbott Catalano and Hawkins 1998 Esbensen Winfree He

and Taylor 2001 Katz et al 2011 Huff 1998 Curry et al 2002 Miller and Brunson 2000

Miller and Decker 2001 Pyrooz et al 2012 Curry et al 2001) and disproportionately affect

neighborhood levels of crime and violence (Block 2000) These findings have been robust in

that they have been repeatedly found regardless of research methodology (ie use of official

data self-report data observational data) (eg Curry 2000 Decker 1996 Deschenes and

Esbensen 1999 Esbensen et al 2001 Gordon et al 2004 Katz Webb Schaefer 2000 Webb

Katz Decker 2006) or research setting (ie North America Europe Asia South America

Australia) (Klein and Maxson 2006) As a consequence it should not be surprising that

policymakers academics and community leaders have focused much of their attention on

developing responses to address community gang problems

Suppression strategies have been the favored public policy response to gangs since the 1980s

(Venkatesh 1999 Spergel et al 1995) Suppression strategies typically rely on focusing

criminal justice resources on gang members through such practices as targeted and enhanced

police patrols intelligence databases vertical prosecution and enhanced sentences for those

convicted (Katz and Webb 2006) Suppression strategies are based on deterrence theory and

are founded on the principal that swift certain and severe penalties for those involved in gangs

and gang activities will necessarily result in fewer individuals joining gangs and will deter people

from engaging in gang crime (Klein 1995) By the early-to-mid 1990s as gang problems

continued to proliferate and policymakers sought alternative gang control strategies gang

prevention programming flourished (Papachristos 2013) Gang prevention programs were

aimed at the general youth population or focused on at-risk youth or neighborhoods Gang

prevention programming was based on the premise that by reducing risk factors and increasing

2

protective factors prevention specialists could inoculate youth from gang membership

(Esbensen 2000) Prevention programming has most frequently come in the form of school-

based prevention programming such as the Gang Resistance Education and Training (GREAT)

program where students are exposed to a curriculum designed to reduce both participation in

gangs and gang crime as well as the SolucionES3 program which engages youth and their

families through integrated community-based and school based programs

While a burgeoning body of literature has developed describing the assumptions issues and

effectiveness of suppression (eg Decker 2003 Decker and Reed 2002 Katz and Webb 2006

McCorkle and Miethe 2002) and prevention strategies (Esbensen and Osgood 1997) much less

attention has been given to examining gang intervention programming This might largely be

the consequence of the absence of such programming over the past 30 to 40 years By the

1980s policymakers no longer believed that social intervention approaches were an effective

strategy to control gangs and gang problems Although gang intervention strategies took many

forms they were based on the assumption that gang membership is the by-product of a socially

deprived community and that the values and norms of gang youth can be influenced and

directed toward those of mainstream society As such intervention programs often focused on

diverting youth from gangs or sought to minimize the consequences of gangs and gang activity

Gang intervention strategies include crisis intervention dispute resolution street-level

counseling and youth outreach (Spergel 1995) However many policymakers community

organizations and academics have argued that such approaches not only did not reduce gang

activity but may have lead to increased group cohesiveness which in turn may have lead to

increased delinquency (Klein 1971 Spergel 1995) More recent research has yielded similar

results For example a number of studies examining replications of Chicago CeaseFireCure

3 SolucionES is a Global Development Alliance in which $20 million is provided by USAID and SolucionES Alliance

members will contribute an additional $22 million dollars during a 66-month period starting July 2012 SolucionES is being implemented by an Alliance of five leading Salvadoran non-profit organizations who have come together to prevent crime and violence The Alliance members are Fundacioacuten Nacional para el Desarrollo (FUNDE) Fundacioacuten Salvadorentildea para la Salud y el Desarrollo Humano (FUSAL) Fundacioacuten Crisaacutelida (known locally as Glasswing) Fundacioacuten Salvadorentildea para el Desarrollo Econoacutemico y Social (FUSADES) and Fundacioacuten Empresarial para el Desarrollo Educativo (FEPADE) Together these five organizations have notable expertise in education health community development economic development research and youth leadership they are bringing their combined synergy and strengths to prevent crime and violence in El Salvador More information on social violence prevention programming is presented in Annex E including a summary of the violence prevention initiatives being implemented by SolucionES

3

Violence which relies heavily on crisis intervention dispute resolution street-level counseling

and youth outreach have found these strategies to either be ineffective or worse increase

levels of violence (Fox et al 2014)

One type of gang intervention gang truces however has received little attention in the

literature This is somewhat surprising given its use and public claims of effectiveness The

purpose of this report is to systematically examine the effectiveness of gang truces Gang truces

have largely been a ldquoblack boxrdquo that is little is known about the frequency of their use

conditions that give rise to them the role of third parties in brokering gang truces the

transformative effects of truces and the effectiveness of gang truces In this report we review

prior research on gang truces and present evidence on the effectiveness of gang truces that

have been implemented in El Salvador Jamaica and Honduras

Theoretical and Policy Rationales for Gang Truces

Much of the concern about gangs over the past twenty years has been their close association

with violence Qualitative and quantitative research has repeatedly found that gangs and gang

members are involved in high levels of serious violent offending Decker (1996) attributes the

gang-violence relationship in part to the collective and normative structure of gangs which

supports and encourages the use of violence both preemptively and in retaliation He further

explains that gang membership encourages the use of violence in retaliation against threats

and attacks whether actual or perceived which results in patterns of inter-gang conflict

characterized by escalating violence Related Cooney (1998) points out that gangs are not all

that different from ldquowarrior societiesrdquo He argues that while gangs are violent the violence

that they engage in is different than that engaged in by non-gang members in that it commonly

takes on a feud-like dimension A perceived slight violation of turf or other disrespectful

action might invoke a shooting which in turn evokes a retaliatory shooting which in turn

results in another retaliatory shooting creating a self-perpetuating cycle of violence

Early research and theoretical work examining gangs and gang conflict suggested that much

violent gang behavior was the function of status management (Thrasher 1927 Whyte 1943

Short and Strodtbeck 1965) Borrowing from the sociology and social psychology literature on

4

impression management theory gang scholars hypothesized that youth place significant

emphasis on image managementmdashseeking to impress their peers and limiting the potential to

embarrass themselves in front of others (Hughes 2005) These early gang theorists postulated

that violence is an instrument used by gangs and gang members to achieve manage and

protect status They contend that gang members often make decisions to become involved in a

violent conflict based on rational processes that weigh the immediate loss or gain of status

within the gang against the relatively small probability of being formally sanctioned by officials

within the criminal justice system (Hughes 2005) A number of researchers have examined the

relationship between status considerations and gang violence and have found the association

to be particularly robust regardless of gender (Campbell 1991 Giordano 1978) ethnicity

(Anderson 1990 1998 Vigil 1988) and location (Jankowski 1991 Spergel 1995) Hughes

(2005) notes that the centrality of status consciousness by gang youth may account for the

existence of the facilitative gang effect that gang scholars have repeatedly observed (ie Gangs

increase the amount of delinquency individuals are involved in beyond that of a group of

delinquent peers (Gordon et al 2004 Thornberry et al 2003)

Another micro-social factor associated with gang violence is group cohesiveness While our

understanding of the relationship between gang cohesiveness and violence is underdeveloped

some scholars have reported a strong relationship between the two In particular Klein and his

associates (1971 1967) and Lucore (1975) have reported that increases in gang cohesiveness

lengthens periods of gang membership and increases members participation in gang crime and

violence Cooney points out that there are strong relational ties between gang members that

necessarily result in increased cohesiveness among members For example gangs are

comprised of neighborhood youth who share common cultural and economic experiences and

have often known one another for lengthy periods of time He also points out that gangs are

groups that have strong self-proclaimed and formalized identities (eg colors symbols names

monikers) and have at least some organizational structure Their sense of group is maintained

by their common understanding of their members and friends and their attachment to their

territory (or neighborhood) Decker (1996) notes that the relatively high level of group

5

cohesiveness exhibited by gangs facilitates both collective behavior and liability (For exception

see Hughes 2013)

Accordingly both micro-social factors (ie status management gang cohesiveness) serve to

augment levels of gang violence and make it difficult for third parties to intervene Violence

within the context of gangs serves as a form of informal social control Gangs and gang

members cannot seek assistance from legitimate institutions of social control to solve conflicts

because they would risk losing status (Anderson 1999) Likewise the collective nature of the

gang not only increases potential offenders and targets of violence but also facilitates at the

group level the need for retributive justice on the behalf of injured members Moreover gang

members desire to impress others with their commitment to the group and use violence to

demonstrate their commitment to their group and to increase their status within the group All

of this results in an increased cycle of gang conflict and violence4

Over the past several decades residents and policymakers have expected the police to address

violent gang conflicts However there are several limitations to the police response to gangs

First as noted above many of the same factors that are associated with violent gang conflict

also limit the effectiveness of the police to have an impact on violent gang conflict Gang

members are not going to contact the police to resolve a conflict because it could result in loss

of status and expose them to the police discovering illegal activities that they are involved in

(Katz 2001 Katz and Webb 2006) Related citizens in neighborhoods with gang problems are

also reluctant to call the police out of fear of gang reprisals (Katz 1997 Webb and Katz 2003)

or because they have a poor perception of the police themselves (Katz Choate and Webb

2002) Third the police response to gangs in most communities is a reactive response to a

specific incident after it has occurred rather than a proactive response to intervene in ongoing

disputes between gangs (Katz and Webb 2003 2006) Most police agencies simply do not have

4 There is anecdotal evidence for these hypotheses Hughes and Short (2005) through field observations and interviews with members of 20 Chicago gangs examined the micro-social context of violent-related incidents Specifically they were interested in the specific social conditions that precipitated violence Their analysis revealed that gang disputes associated with retaliation escalated into violence significantly more often (ie 55 of the time) than disputes associated with normative order violations or an identity attacks which only resulted in violence roughly 33 percent of time

6

the intelligence networks required to intervene in gang conflicts until after they have risen to

relatively high levels (Katz 2003) Fourth police suppression strategies have been linked to

increasing gang cohesiveness and possible increases in gang crime (Klein 1995)

Some policymakers and community activists have proffered that an alternative to reliance on

formal mechanisms of social control such as the police is the gang truce (Spergel 1995) The

goal of a gang truce is to reduce or even eliminate violent conflict between those gangs that are

warring with one another As such compared with other strategies that often seek to reduce

general levels of crime a gang truce has the very specific goal of reducing violence between

two or more gangs that are in conflict with one another Unfortunately the literature provides

little guidance on the theoretical assumptions of why a gang truce should reduce inter-gang

violence Henderson and Leng (1999) hypothesize that at the root of gang truces is the notion

that they involve the renegotiation of existing norms within and between gangs The authors

claim that as a violent dispute escalates between gangs leaders and members are placed in the

situation of appearing weak to both members of their own gang and to members of the rival

gang if they do not respond with the appropriate amount of force or if they were to suggest a

peaceful resolution to the dispute Accordingly Henderson and Leng argue that as the cycle of

violence escalates between two or more gangs behavioral norms shift toward the increased

valuation of violence to resolve the conflict because it is the only option readily available to

them A gang truce on the other hand which is often mediated by a third-party is believed to

break the cycle of violence by providing the gangs involved in the dispute with a cooling-off

period (Spergel 1995) In the interim period new norms of expected behavior within and

between gangs are established In other words the cooling-off period is believed to recalibrate

norms of behavior that are more consistent with the security interests of the gang and its

members (Henderson and Leng 1999)

For many of the reasons stated above a number of communities have participated in gang

truces Gang truces have been observed in the United States Central America and the

Caribbean and gang leaders in other nations are considering its implantation (Fahah 2012)

Unfortunately little is known about gang truces We do not know when they come into

consideration how they are implemented whether they decrease increase or even have a

7

significant impact on violence In the below section we systematically review the existing body

of literature on gang truces Appendix A describes the methodology we employed to conduct

the systematic review

Prior Research on Gang Truces

Little research to date has examined the effectiveness of gang truces and of the research that

has been conducted has been restricted to the field of public health The most celebrated

truce to date was Philadelphiarsquos House of UMOJA 1974 gang truce In preparation of the truce

gang leaders in prison were called upon for their support along with key community leaders

from churches businesses schools and the police When UMOJA called for the gang summit it

was reported that 500 gang members and 75 of Philadelphia gangs attended By the

conclusion of the gang summit a 60-day truce was announced which reportedly resulted in no

gang members being killed over the 60-day period However no systematic evaluation of the

gang truce was conducted (Woodson 1981)

The first attempt to evaluate a gang truce to our knowledge was conducted by Cotton (1992)

who examined the results of a gang truce in South Central Los Angeles between the Crips and

Bloods Data provided by the police department indicated that over the 6-week period when

the truce took place drive-by shootings decreased by 48 compared to the same 6-week

period the prior year decreasing from 162 to 85 Likewise gang-related homicides dropped by

62 from 26 to 105

Ordog et al (1993 1995) examined the effects of a gang truce in Los Angeles using emergency

room admissions data Specifically the authors examined changes in the daily and monthly

number of gun shoot wound (GSW) emergency room admissions before during and after a

gang truce The catchment area for the emergency room was 100 square miles The authors

noted that while they were able to clearly identify the date the gang truce began because of

the media coverage that it received there was no specified date that the truce ended and as a

5 The authors did not report whether the drop in gang-related homicides was citywide or in the South Central Los Angeles neighborhood where the gang truce took place

8

consequence after 12 weeks the authors regarded it as called off for evaluation purposes

because it was no longer being discussed in the media and gun shot wounds began to increase

Student t-tests were used to examine changes before during and after the gang truce Ordog

et al (1995) reported that there were approximately 7 GSW admissions per day the 12-months

preceding the truce compared to 45 GSW admissions per day during the gang truce and 126

GSW admissions per day in the 11-months following the gang truce The authors concluded

that their analysis ldquoclearly showed that the institution of a gang truce had reduced the number

of GSW victims seen in an hellip inner city Level I trauma centerrdquo (p 419)6

However it is important to point out that while the gang truce in Los Angeles did decrease

homicides by about 35 percent for the first three months it then doubled in months four

through eleven compared with the pre-truce period - a pattern that the authors did not

discuss Similar findings were reported in Trinidad and Tobago (Maguire Katz and Wilson

2013) where it was determined that homicides declined for a brief period of time (again for

about three months) but then increased substantially over the long term (12 months) These

results suggest that gang truces may produce short-term benefits yet result in long-term

adverse consequences

While the research examining gang truces shows their potential promise and their potential for

greater harm we believe that the findings should be viewed with caution for two reasons

First some of the implicit theoretical assumptions that gang truces are built on may not be

accurate Prior research on gangs suggests that they have limited organizational structure

(Decker Katz and Webb 2008 Decker Bynum Weisel 1998) and have few formal

mechanisms to influence member behavior For example most gangs do not have formal

leaders do not require members to pay dues and members do not make contributions to the

gang for the purpose of developing the gangs infrastructure (ie guns housing etc) A number

of academics have pointed out that ldquosophisticated gang organizations is still largely a product of

the self- or organizational-interested musings of gang leaders certain police officials academic

6 Similar results were presented by Ordog et al 1993 using the same data but analyzing it over a shorter period of time and aggregating the data by month instead of by day

9

researchers media reporters based on very limited hard datardquo (Spergel 1995 79-80)

Therefore even if a truce was successfully negotiated between members of gangs in conflict

much of the academic research suggests that gang leaders do not have enough control over

members to enforce a truce Related prior research on gangs and gang members suggests that

they are primarily comprised of young people with few ldquodiplomaticrdquo skills (Henderson and Leng

1999) The very same organizational and normative features of gangs that result in gang

violence (ie often bastions of young minority male street youth) are those same features that

most likely limit the likelihood of a gang having the capacity to abide by a gang truce In sum

implicit in the theoretical assumptions of a gang truce is that gangs have the organizational and

cultural capacity to create and maintain (at least for a short while) a truce which is inconsistent

with much prior research on the organizational characteristics of gangs

The second reason we believe that prior evaluations of gang truces should be viewed with

skepticism is that they have relied on relatively weak evaluation designs For example prior

studies examining gang truces have not incorporated the use of comparison areas or control

groups The causes of reductions in gang violence found in previous evaluations might be

many For example the Los Angeles riots took place just before the gang truce that Ordog et al

evaluated Zinzun (1997) reported that gang culture and violence changed briefly but abruptly

following the riots because gangs and gang members in part redirected their anger and focus

toward the police As such the decline in GSW admissions may have been the consequence of

an overall city-wide decline in gang violence in the wake of the LA riots Related prior research

has relied on fairly broadly defined outcome measures such as GSW admissions and general

levels of drive-by shootings Such outcome measures lack specificity in terms of attributing

gang violence to the specific gangs involved in the gang truce Evaluation designs used in

previous studies have also lacked specificity in terms of the catchment area where violent

activity was labeled as gang related For example Ordog et alrsquos (1993 1995) outcome measure

included all emergency room admissions for GSW However the emergency room received

patients from a 100 mile square area surrounding the hospital an area that was most likely

much larger than the gangsrsquo territories involved in the truce

10

Additionally prior research examining gang truces has not examined the processes involved in

the creation of the gang truce No context has been provided in terms of the factors that lead

to the gang truces whether the gangs were pushed or pulled into truces whether outside

parties helped mediate the truces or whether on-going mediation was required to maintain the

truce In other words we still do not know the processes related to the formation of a

successful gang truce Instead prior research and evaluation has treated the gang truce much

like a black box where it is described in very general terms but its details are not revealed

Likewise prior evaluations have not examined the effects of a gang truce beyond its impact on

gun violence A number of prior researchers have argued that mediating such activities serves

to legitimize gang leaders increases gang identity for members and results in greater group

cohesion (Haskell and Yablonsky 1982 Klein 1995) As such we do not know if there are any

unintended consequences resulting from gang truces due to the methodologies chosen for

prior research and evaluation Finally prior research on gang truces has not relied on more

sophisticated statistical analysis that has the capacity to not only discern overall trends but also

cyclic patterns outliers and turning points

THE PRESENT STUDY

The present study seeks to understand the effectiveness of the gang truce when negotiations

occur with or between gangs and with government institutions or civil society organizations

We have briefly reviewed the theoretical and policy rationales for gang truces and the

literature and prior research on this form of gang intervention Next we present three case

studies one from El Salvador another from Jamaica and another from Honduras Each offers

evidence about their impact on violence and lessons learned that are unique to their particular

circumstances In conclusion the case studies are considered together and we present their

collective policy implications

11

Case Study A Gang Trucemdash The Salvadorian Experience

Over the past two years the Salvadoran gang truce brokered by an array of local policymakers

and international development organizations has attracted national and international

attention Violence in El Salvador has been at an all time high with a rate of 692 homicides per

100000 population in 2011 (UNODC 2014) making El Salvador one of the most violent nations

in the world (Parkinson 2014) Since 1992 government and non-government actors have been

responding to the rising tide of gang violence using traditional suppression strategies such as

increased policing legislative changes and more severe prison sentences These traditional

mechanisms of formal social control however were proving ineffective if not counter-

productive (Perez 2003)

In response to the inadequacy of traditional strategies stakeholders altered their course in an

effort to radically reduce gang violence in the nation Members of the Funes administration led

a group of negotiators comprised of the Catholic Church a former congressman and the

Organization of the American States (OAS) to help frame the conditions for a possible truce

between the MS13 and 18th Street gangs (Umana de Leon and Tager 2014) In March 2012 a

truce was reached The goal of the gang truce was to reduce violence specifically gang-involved

homicides Included in the terms in exchange for the gangs acting to reduce homicides certain

incarcerated gang members were to be transferred to lower security prisons to receive

sentence reductions and special visitation privileges and to be permitted more communication

with those outside of the prison for the purpose of conducting crisis interventions to mitigate

the violence (Salanegra 2012)7 The gang leaders also agreed to no longer recruit children into

their gangs reduce violence against women give up a small number of guns and continue to

participate in negotiations (Seelke 2014 11-12)

7 It is important to note that there were other discussions that took place during the mediations that included as a local publication noted ldquoProhibition Act Gang send the army to barracks end the police operations in controlled areas by gangs repeal of the figure of the witness criteriado (with criteria of opportunity or ldquowitness under a plea agreementrdquo) and a number of improvements in the quality of life of the inmates and its familiesrdquo Source httpwwwsalanegraelfaronetes201209cronicas9612

12

The present case study seeks to understand the role of negotiations with or between gangs and

government institutions or civil society organizations and their impact on violence in El

Salvador The specific objectives of this case study were to (1) Identify and document the

negotiation processes with or between gangs This includes but will not be limited to

identifying the actors involved in the negotiations the goal(s) of the negotiations and

strategies employed to carry out negotiations (2) Identify the impact of negotiations with

andor between gangs on violence and other forms of criminality and (3) Present conclusions

about the strengths and weaknesses of negotiations and their potential for positively or

negatively reducing crime in Salvadoran society

The Salvadorian Gang Truce

While the origins of the Salvadorian gang truce are somewhat unclear some accounts suggest

that in the fall 2011 the Minister of Security (then David Munguia Payes) mentioned the idea of

starting a dialogue between the gangs to Raul Mijango (a former congressman) Shortly

thereafter Mijango spoke to Monseigneur Fabio Colindres the military chaplin about the idea

They started to communicate with gang leaders in January 2012 From the beginning the

Minister of Security stated that he would not personally communicate with gang leaders and

noted that the official position of the government was not to negotiate with offenders But he

did allude to the fact that the gang truce was part of a new strategy to address the nationrsquos

gang problem (Archibold 2012) The Salvadorian gang truce was multi-dimensional involving a

varying number of actors communication styles and tactics In the section below we discuss

the parties involved in the truce the negotiation process and the strategies used by the gangs

and government in furtherance of the truce

Parties involved in the truce There were a series of persons groups and other entities who

directly or indirectly participated in the negotiation process whether they were negotiators or

collaborators From the beginning negotiations between the MS13 and 18th Street gangs were

considered confidential and were held in strict confidence between the gangs and the

negotiators Beginning in early 2012 the first two mediators to serve were a former

congressman (Rauacutel Mijango) and a priest (Fabio Colindres) These individuals were able to gain

13

the trust of the gangs and also had open communications with the government and

community

As the process moved forward however church leadership changed its initial position and

withdrew its support of the truce This left the priest to participate as a private individual no

longer serving as a representative of the church The mediators had some early success

however as the mediators changed the lines of communication became less clear and less

reliable Still they played an important role in the process Finding individuals whom the gangs

would trust and who could speak for government and community institutions was challenging

Further as government ministers and presidents changed it became less clear whether or not

the government was participating in the dialogue between various parties much of the

confusion in negotiations was related to understanding the governmentrsquos official role in the

process

In the beginning the negotiating parties included the two mediators and MS13 and 18th Street

the two predominant gangs in El Salvador MS13 is the largest gang in the country with about

250 cliques throughout the nation For the most part its first- and second-generation leaders

were in prison There are municipalities in which only MS13 operates and therefore its

members never clashed with 18th Street adversaries MS13 members are largely involved in

extortion violence and intimidation of the public The MS13 leadership has been shown to

have greater control of its organization than most its members have been more disciplined

and despite its size the gang seems to have sufficient and effective communication

mechanisms between cliques and its leadership (ldquoranflardquo or ldquoruedardquo)

The 18 Street gang is somewhat smaller than MS13 Long before the truce process began

factions existed within 18th street which had become deeply involved in internal conflicts over

leadership structure Those differences had become more pronounced over time leading to the

establishment of two movements known as the ldquoSurentildeosrdquo (Southerners) and the

ldquoRevolucionariosrdquo (Revolutionaries) that participated in the peace process but under different

14

representation Despite their differences however for the negotiation process the two

fractions of 18th street established similar demands between themselves and with MS138

As noted above the governmentrsquos involvement in the truce process was never clearly

articulated and it varied with ministerial changes The Funes Administration specifically the

Ministry of Justice and Public Security (MJPS) had been engaged in the negotiation process

early on In 2013 however with the change in MJPS leadership this also changed Tensions

grew over the role of the government in the truce negotiations According to the mediators the

new Minister of Security wanted to dismantle the process and for this reason he blocked the

mediatorsrsquo entry to the prisons where they had been meeting with gang leaders The change in

government leadership brought changes in strategy As with all negotiations mutual trust was

the cornerstone When the personnel and their agendas changed so did the trust in the peace

process

Late in 2012 as negotiations began to focus on the relationship between the gangs and the

community other organizations were called upon to help facilitate this interaction A technical

committee was created to oversee progress towards the violence reduction goal this was

announced at the end of August 2012 The committee moved forward with support from the

Organization of American States (OAS) It coordinated actions intended to improve the living

conditions of incarcerated persons to reintegrate into society those who had served their time

in prison and to prevent violence and provide assistance to victims of violence The committee

operated as a mechanism for coordinating the negotiation process with technical support from

the United Nations Development Program (UNDP)‒El Salvador Interviews with experts

indicated that Fundacioacuten Humanitaria with support from several organizations like Interpeace

and others had also supported the operations of the mediators and served as an important

point of reference for the gangs

In each territory where agreements were made government representatives from some

municipalities were invited to participate in the process as key implementation stakeholders

Between December 2012 and January 2013 the municipalities of Ilopango Santa Tecla

8 httpsalanegraelfaronetes201110cronicas5645

15

Sonsonate and Quezaltepeque became among the first to become part of what later became

known as the Sacred Municipalities (aka violence free municipalities) By November 2013

several other municipalities had become part of this group La Libertad Apopa San Vicente

Puerto El Triunfo Nueva Concepcioacuten Ciudad Delgado and Zacatecoluca

The gangs and the government offered several incentives to one another The gangs offered a

number of actions to reduce the incidence of crime (ie cease of hostilities zero homicides

stop extortions robberies and recruitment of children and youths into gangs peace at the

schools) and the government offered to guide public investments in social policies prioritizing

participating municipalities

The truce process was the product of a convergence of vested interests from different sectors

Several earlier attempts at negotiations had failed likely because of a lack of perceived

authenticity on the part of the actors All parties involved in this truce process came to the

table with well-defined group interests and concerns All sides were experiencing fatigue from

the long history of violence Years of fighting deaths and violence had led the mostly

incarcerated first generation those who were still living and who continued to serve as gang

leaders to seek new ways of exercising social power and influence in prison and the

community

The negotiation process In El Salvador the negotiation process between gangs formally began

on March 8 2012 by a decision of the Government to favor the transfer of gang leaders

imprisoned out of the countryrsquos only maximum security prison located in the municipality of

Zacatecoluca That decision was made by the Minister of Justice and Public Security as part of

an agreement between a team of mediators who served as advisors to the Ministry and gang

leaders primarily from MS13 and 18th Street The negotiations were referred to as a ldquotruce

between gangsrdquo and were characterized by the mediators as ldquoa peace processrdquo or a ldquocease of

hostilities between gangsrdquo

Publicly the gang truce was known as an agreement between gangs and not between gangs

and the government because of fear of how those outside the negotiations might interpret the

governmentrsquos actions From the beginning therefore the government never fully

16

acknowledged its participation in the peace process (through representatives) with the gangs

and that reticence was a major impediment throughout the process This was largely because

of the difficulty in communication between the multiple parties which was done through

mediators and necessarily resulted in additional logistical complexity throughout the

negotiations Regardless it is important to recognize that the government played a decisive

and central role in the origin facilitation and promotion of the peace process

In terms of processes interviews with gang leaders indicated that they would communicate

their ldquodemandsrdquo to the mediators who in turn would communicate the gang leaderrsquos demands

to the government The same process was used for the government to communicate with the

gang leaders After an agreement was reached the parties would make proclamations that

emphasized the agreed upon terms of the negotiations These often took the form of public

statements made through local newspapers For example there were more than twenty public

statements made by the gangs where they clearly articulated that their intention was to

reduce the harm that their acts of violence were causing themselves and the communities (For

a detailed discussion of the public statements see Appendix B) In these public statements the

gangs acknowledged the need for their groups to change Both the mediators and the gangs

made clear that the gangs expected the government to respect and respond to their demands

given the good-faith actions that the gangs had already taken

The government on the other hand did not make any formal public statements about their

end of the agreements reached through the negotiations From the outside it appeared that the

government did not develop a formal or consistent policy regarding the truce and instead

assumed that they could achieve a reduction in homicides through the negotiation of increased

prison privileges for gang leaders At the close of the Funes administration the peace process

and negotiations initiated within its framework had reached a plateau There they remained

waiting for the new government to step in and take action as President-elect Salvador Saacutenchez

had announced his support for the truce during the electoral process Sanchez however

withdrew that support when he assumed the presidency in June 2014 Indeed under the

recently elected Sanchez administration the government changed course and became less

17

willing to engage the gangs believing that the government should not negotiate with criminal

organizations

Indeed it should be noted that responses resulting in increased gang influence have been a

concern over the past several decades--from governmental neglect of the problem in the

1990s through the period of the mano dura and super mano dura of the early 1990s to

2000 and to the present gang truce Many have asserted this claim as truth (ie some

responses will increase the influence of gangs) especially in public discourse This study

however does not directly address this point which may or may not be valid9

Strategies employed during negotiations between the gangs and the government Over the

course of the negotiations between the gangs and the gangs and the government at least two

strategies were employed At the beginning of the negotiations strategies were implemented

similar to that of the historic Salvadorian peace accords Each of the parties with the support of

mediators reached agreements and achieved a resolution to their conflict This model

although useful overlooked an important difference between the current violence in El

Salvador and the civil war of the 1980s In a conflict of civilians with the State the legitimacy of

the ldquoadversariesrdquo arises by the need of the state to recover the rule of law It is certainly

possible to reach peace agreements that resolve conflict between gangs but in the current

case with regard to the government there was no legal or policy justification for executing the

truce process On the contrarymdashgovernment negotiations with a criminal group are relatively

rare in modern democratic societies The lack of a legal or policy framework to work from

limited the governmentrsquos ability to be transparent in its response to gangs and may have

undermined its legitimacy with the public

The second strategy employed by the parties was based on reciprocation and cooperation

between actors Early on it was determined by both sides that if one of the parties abandoned

the peace process or did not hold up its end of an agreement the other party would no longer

participate in the negotiation process (Axelrod 1986) In retrospect during the course of the

peace process the gangs implemented agreed-upon terms and positioned themselves as valid

9 We would like to thank one anonymous reviewer for bringing this issue to our attention

18

partners with the government able to negotiate for what is needed by the government in

exchange for what the gang needed10 For example in public statements the gangs insisted

that they had made a goodwill gesture when they declared a unilateral truce and stopped their

involvement in violence The government responded by transferring incarcerated gang leaders

from a maximum-security prison to ordinary criminal prisons11 Next the gangs offered to

disarm the government responded with increased flexibility on control measures at those

prisons This process continued until the demands from both sides grew in a direction that

challenged each sidersquos capacity and willingness to follow through

For instance mediators through the government asked the gangs to end the extortion of

businesses and individuals which are the primary means of subsistence for Salvadorian gangs

The gangs asked that the government reciprocate by eliminating the ldquowitness under a plea

agreementrdquo from the criminal procedural law which is one of the main weapon laws used for

sentencing gang members for complex crimes such as extortion Each of these requests were

more than the parties could carry out Exhibit 1 outlines the different ldquooffersrdquo and ldquodemandsrdquo

made by each party The offers and demands are divided into those that were believed to be

relatively simple and were ldquoofferedrdquo by the party on its own volition and those that were

believed to be more complex and were demanded by the other party

Exhibit 1 Simple demands and concessions made by gangs and the government

Gang concessions Government concessions

Reduce the homicides by more than 50 Transfer the leaders to prisons with lesser security

Hand over 500 firearms to the authorities Allow visits of the children of the gang members

Reduce violent actions at schools Allow night intimate visits

Stop killings at the prisons Allow entry of fast foods

Do not murder custodians police soldiers or their family members

Facilitate the entry into the prison of gang members let out of prison

Give opportunities to some gang members to withdraw

Suspend the searches by the armed forces on persons at the prisons

10 Source httpwwwsalanegraelfaronetes201209cronicas9612 11 This was useful to facilitate the coordination and the communication between the gang members in prisons and in the streets

19

Make it more flexible for new social plans by government to be established in the communities controlled by gangs

Improve the conditions and health assistance at the prisons (eg place tvrsquos to improve mental health of inmates)

Do not burn buses do not kill bus drivers or collecting agents in buses

Allow the Red Cross and journalists to enter into the prisons

The second agenda presented in Exhibit 2 shows the more complex demands that were

not offered by one side or the other but rather were demands placed on the other party These

demands were such that they required a higher level of authority to negotiate in order to

implement the demanded action These demands were considered critical for the peace

process to continue and the delayed responses on both sides stalled the progress of the truce

and led to the parties questioning the legitimacy of the other side all these demands have yet

to be attained The mediators recognized at the end of 2012 that some of the slow progress

was related to the difficulty of making the transition from offers that could be executed

relatively simply to the more complex demands made by each party One of the gang leaders

summarized the issue ldquoWe are not looking for television sets while all of our people continue

living like shit we are not going to try to do everything that is in our power to decrease as much

violence as possible for one television set There are things we are very clear about this [points

to a television set in the room] is a right that the law grantsrdquo

Exhibit 2 Complex demands and concessions made by gangs and the government

Demands by Gangs Demands by Government

Eliminate the figure of witness under a plea agreement from the Criminal Law

Stop homicides and extortions indefinitely

Create international commission to investigate the human rights violation cases of the gang members and their family members on the part of the State (PNC and FA)

Share the information regarding the whereabouts of the brunt of extortion money whether in country or abroad

Institutionalize external and professional surveillance regarding the behavior of the PNC as regards investigations and gang member arrests

Progressive dismantling nation-wide of the clique structures and turfs

Army should stay in its barracks and definitely suspend their participation in public security tasks

Permanently suspend the orders to murder State security and justice agents and their family members (police soldiers judges prosecutors)

20

Clearing of judges prosecutors and police involved in corruption cases against youths in conflict with the law

Permanently suspend the murders extortions and harm caused to public transportation resources and their workers

Maintained a sustained improvement on the conditions of the prisons

Permanently suspend the recruitment of children and youths and hostilities to educational centers

Guarantee working opportunities for the gang members and their family members through specific programs at the municipalities

Share the information about providers of drugs and illegal arms

Television sets were one of the concessions made by the government from the simple list The

gang leader recognized that their intentions in the negotiations extended beyond improving

prison conditions however the government did not have support nor established mechanisms

for carrying out tasks derived from the more demanding list Further as the government

transitioned to new presidential leadership many of the more simple concessions had already

been made and the new government would have to address more complex demands if

negotiations were to continue

The Salvadorian gang truce is remarkable for several reasons First a number of policymakers

and researchers have claimed that the truce saved a large number of lives and was perhaps the

most successful gang truce in the Western Hemisphere Second the Salvadorian gang truce is

somewhat unique in that it might have lasted substantially longer than any other successfully

negotiated gang truce Understanding the temporal impact of the truce is important to

understanding its future potential Third the Salvadorian gang truce is important because a

number of other counties have sought to replicate it For example following the perceived

(and perhaps real) success of the Salvadorian gang truce the nations of Honduras Belize and

Guatemala instituted or tried to institute a similar type of truce Understanding the impact of

the Salvadorian gang truce will further help us understand the potential impact of such a

process on violence in these other nations

Methods

We examined the impact of the gang truce by merging four separate data sets First data from

the 2007 El Salvador Population and Housing Census provided municipal level measures of

21

percent unemployed percent male aged 10-29 percent female headed households ethnic

composition in-migration income percent urban percent households rented and percent

who had a high school education These data were obtained directly from El Salvadors Ministry

of Economy Second municipal level crime data (ie homicide extortion rape theft robbery

and auto theftrobbery) were provided by the National Civil Police (PNC) by month and year for

the period between January 1 2010 and June 30 2014 Disappearance data was also provided

by the police aggregated by year and municipality Third police gang intelligence data was

provided by the National Civil Police (PNC) Specifically we received the number of police

recorded MS 13 and 18th Street gang members by municipality in 2011 the latest year for

which these data were made available Last we acquired 2011 prison gang intelligence data on

the number of imprisoned MS 13 and 18th Street members by municipality from the

Salvadorian National Bureau of Prisons All four data sources were linked together using a

unique municipality identification number and month Collectively they provide an

opportunity to examine the impact of the gang truce in El Salvador and understand several

competing explanations for any changes that might have occurred

The geographic unit of analysis for the present study is the municipality According to the 2007

El Salvador Population and Housing Census there were 14 departments divided into 262

municipalities (the equivalent of a county in the USA) in El Salvador Of the 262 municipalities

ten (38) of the municipalities were eliminated from the analysis because of missing data

Measures

Dependent variables The dependent variable in the present study is the monthly number of

homicides in each El Salvadoran municipality expressed as the number of homicides in

municipality i at time t Homicide data were collected from January 1 2010 through June

30 2014 for a total of 54 months This includes data for a period of 26 months prior to the gang

truce and 28 months following the implementation of the gang truce As shown in Exhibit 3

there were a total of 14148 homicides over the study period with each municipality averaging

371 homicides per month

yit

22

Exhibit 3 Descriptive statistics

n mean sd

Urban Opportunity Factor 252 111 95

male 10-29 years old 252 1960 171

female-headed household 252 3433 499

unemployed 252 1177 686

Racialethnic heterogeneity 252 17 13

of prisoners MS13 252 1579 3960

of prisoners 18th Street 252 1268 4096

of prisoners MS13_spatial weight 252 1579 3960

of prisoners 18th Street _spatial weight 252 1268 4096

MS13 on the street 252 4146 10646

18th Street on the street 252 2513 8622

Violence free municipality 252 04 20

Truce 14148 371 783

Monthly homicide rate per 1000000 14148 51 49

Independent variables Several independent variables were included in our model Our

measure for the intervention is a dummy variable in which the value of 0 is used to represent

the pre-intervention months and 1 represents the post-intervention months We also

included a dummy variable for each community that participated in the violence-free

municipality initiative (sanctuary municipalities) to address issues of external validity (eg

multiple treatment inference) In other words municipalities that participated in the

violence-free initiative might have experienced or participated in something that either

enhanced or detracted from the impact of the larger gang truce The initiative called by some

the second phase of the gang truce took place about 8 to 9 months after the original gang

truce was negotiated in March 2012 In participating sanctuary municipalities gang members

agreed to stop violence and crime in exchange for a reduction in police operations and night

raids12 The municipalities that participated in the initiative included Santa Tecla

Quezaltepeque La Libertad Ilopango Ciudad Delgado Apopa Sonsonate San Vicente

Zacateatecoluca Puerto El Triunfo and Nueva Concepcioacuten

12 See httpwwwcispesorgblogviolence-free-cities-inaugurated-as-second-phase-of-gang-truce

Tt

Si

23

We included several independent variables related to the presence of gangs by gang for each

municipality Specifically we included count variables on the number of MS13 and 18th Street

gang members who were on the streets and in prison by municipality As noted before the

gang truce was made between a relatively small number of imprisoned gang leaders from both

MS13 and both fractions of 18th Street who agreed to stop street-level gang violence on the

condition that they would be transferred to lower security prisons and granted special

privileges

We wanted to understand the impact of the truce in the context of gang presence

Municipalities with high numbers of MS13 and 18th Street members whether they were in

prison or on the streets should have experienced a greater reduction in homicides because of a

greater span of control over these communities Accordingly municipalities with low numbers

of MS13 and 18th Street members should have experienced a lesser reduction in homicides

because of a limited span of control over these communities As Exhibit 3 (above) shows on

average municipalities had about sixteen MS13 and thirteen 18th Street members in prison

and forty-one MS13 and twenty-five 18th Street members on the street We examined whether

gang members in El Salvador were randomly distributed and found that there was spatial

clustering in the number of MS13 and 18th Street members who were in prison In order to

control for the clustering we included the spatial lag (weight) of the MS13 and 18th Street

members in prison

Control variables A series of control variables known through prior research to be related to

violence in communities were added using the census data We began with municipal-level

measures of percent unemployed percent male aged 10-29 percent female-headed

households ethnic composition in-migration income percent urban percent households

rented and percent with a high school education Initial analysis found that five community

variables were highly correlated and loaded on the same factor As seen in Exhibit 4 these

included the percentage of persons who had moved there from another municipality (in-

migration) average income in the municipality percent of the population living in an urban

area percent of houses that are rented and percent of residents with at least a high school

education We labeled this factor Urban Opportunity

24

Exhibit 4 Factor loadings for urban opportunity

Variables Factor

loadings

In Migration Percent of population moved in from another municipality 0638

Income Average monthly income per household (colones) 0886

Urban Percent of population that is urban 0845

Rented Percent of households rented 0761

Education Percent of residents who have at least a high school education 0742

Percent of variance 6071

Eigenvalue 341

Extraction method Principal Axis Factoring

We also calculated a measure of ethnic-heterogeneity from relevant census data Ethnic

heterogeneity which varies from 0 to 1 was calculated by taking one minus the squared

proportions of the population in each ethnic group (White Mestizo other) As with the

percentages of male population that is 10-29 years old female-headed households and

unemployed persons ethnic heterogeneity has consistently been associated with violence in

general (Kubrin and Weitzer 2003) and with gang violence in particular (Katz Maguire and

Roncek 2002 Rosenfeld Bray and Egley 1999)

Analytic Strategy

National-level analysis Our analysis began with a simple t-test of the mean number of

homicides at the national level before and after the truce providing the most basic omnibus

test of an effect We then presented our time series model to estimate the effect net of

seasonality and temporal trends This analysis included a set of simple ARMA (autoregressive

and moving average) models with two lag periods and one period of a moving average Initial

analysis of the number of homicides by month indicated that partial temporal autocorrelations

existed for two lags The first model used only data prior to the truce This model included a

linear time trend variable and dichotomous variables for each month (except January) We used

this model to forecast the expected number of homicides for the truce period This series of

analyses was for illustrative purposes only as the number of data points used in the forecast

25

was too small to provide meaningful confidence intervals The second ARMA model employed

all data from January 2010 through June 2014 and included a dichotomous indicator for the

truce period The effect of this variable was the average change in the number of homicides

net of seasonality (months) and temporal trends

Municipality multilevel models The analysis at the municipal level presented four challenges to

a typical regression model First since our outcome has a highly skewed distribution with

varying levels of exposure (population) we employed a generalized linear model to capture the

correct distribution In this case we employed a negative binomial distribution rather than a

Poisson model due to the over-dispersion in our outcome created by analyzing monthly

municipal datamdashthat is there were several months and municipalities where no or very few

homicides occurred

The second challenge was our need to measure the between-municipality variation of pre-truce

homicide rates and the program effect A fixed-effects model would have been inappropriate

because it would not have allowed us to estimate these variance components Thus we

employed a multilevel random effects generalized linear model

The third challenge was temporal autocorrelation Analysis at the national level indicated

partial autocorrelations in the first and second lags Although a generalized model does not

allow for auto-correlated residuals we addressed this by entering in the model two lags of the

homicide rate

The fourth and final challenge to this estimation was spatial autocorrelation An examination of

the Moranrsquos I and autocorrelation coefficients (exhibit not shown) indicated a low level of

autocorrelation for each month but many months were still statistically significant Thus we

estimated spatial lags of the homicide rate and entered them into the model

Accordingly we analyzed the data using random effects negative binomial models regressing

the rate of homicide on various months on predictors with both temporal and spatial lags For

clarity we present the model using HLM (Bryk et al 1996) notation At the first level (time) we

estimated the log number of homicides using a negative binomial distribution with predictors

26

that included the truce period calendar month linear month trend a

one period lag of the homicide rate a two-period lag of the homicide rate a

spatial lag of the homicide rate and an over-dispersion parameter which has a

Gamma distribution of Thus the final level 1 model is

At the municipality level (level 2) the intercept is a function of the presence of the

Violence Free Municipality program the log number of MS prisoners the log

number of 18th Street prisoners the log of the spatial lag number of MS prisoners

the log of the spatial lag of the number of 18th Street prisoners the log of the

number of MS members the log of the number of 18th Street prisoners the

control variables detailed above and the random effect Thus the level 2 model for the

intercept is

Also at the municipality level (level 2) the effect of truce is a function of SAFE the

log number of MS prisoners the log number of 18th Street prisoners the log of

the spatial lag number of MS prisoners the log of the spatial lag of the number of 18th

Street prisoners the log of the number of MS members the log of the number

of 18th Street prisoners and the random effect Thus the level 2 model for the

truce effect is

TRUCEt mt TIMEt

HRt-1i HRt-2i

HRti vti

Gamma1

aa

aelig

egraveccedilouml

oslashdivide

mti = expb0 i + b1iTRUCEt + lmmt

m=1

11

aring + l12TIMEt

+l13HRt-1i + l14HRt-2i + l15HRti + vti

eacute

euml

ecircecircecirc

ugrave

ucirc

uacuteuacuteuacute

exp vti[ ] ~ Gamma1

aa

aelig

egraveccedilouml

oslashdivide

b0 i

SAFEi PMSi

P18thi

PMSi P18thi

MMSi M18thi

Ci u0 i

b0 i = g 00 +g 01SAFEi +g 02 ln PMSi +1( ) +g 03 ln P18thi +1( ) +g 04 ln PMSi +1( )

+g 05 ln P18thi +1( ) +g 06 ln MMSi +1( ) +g 07 ln M18thi +1( ) + pC

C

aring Ci + u0i

b1i SAFEi

PMSi P18thi

PMSi

P18thi MMSi

M18thi u1i

27

The key parameters of our analysis are in this expression The truce effect is dependent on

the log number of MS prisoners among other characteristics Thus the percent

reduction in homicides can be calculated from this model using the following formula

which we plot below (see Quantitative Findings below)

To answer questions about the variation in the effects of the gang negotiations we predicted

best linear unbiased predictions (BLUP) of the Truce random effect which estimates the

between-community variance of the immediate effect of the truce Since we assumed that the

interventions were not evenly spread throughout communities in this study the mixed model is

appropriate to address this issue We conducted this analysis for homicide using a Stata

generalized linear mixed model using full maximum likelihood and an identity covariance matrix

of random effects

Findings

Exhibits 5 and 6 present our findings related to the number of police recorded homicides in El

Salvador from January 2010 through June 2014 The trends in the data and the results of the t-

test show that in the 26 months prior to the gang truce there were on average about 354

homicides per month compared with about 218 homicides per month in the 28 months

following the gang truce Exhibit 7 shows that the gang truce resulted in 5501 fewer homicides

than otherwise would have occurred

b1i = g 10 +g 11SAFEi +g 12 ln PMSi +1( ) +g 13 ln P18thi +1( ) +g 14 ln PMSi +1( )

+g 15 ln P18thi +1( ) +g 16 ln MMSi +1( ) +g 17 ln M18thi +1( ) + u1i

b1i

ln PMSi +1( )

1- exp g 10 + ln PMSi +1( )g 12( )eacute

eumlugraveucirc( )acute100

u1i

28

Exhibit 6 Number of homicides in El Salvador by month with ARIMA Model Fit

Exhibit 5 Change in homicides from pre-truce to post-truce

Obs Mean SD 95 confidence interval

Pre-truce period 26 35442 4200 33745 37138

Post-truce period 28 21835 6361 19369 24302

Change -13606 10638 16574

=p lt 05

29

Exhibit 7 Forecast of homicides without gang truce

Next as seen in Exhibit 8 we examined the spatial distribution of the change in the homicide

rate after the implementation of the gang truce The analysis showed that of the 252 analyzed

municipalities 243 (93) experienced a decrease in homicides however within these

municipalities there were wide variations in the degree of the decline For example of the 243

municipalities that experienced a decrease the decrease in the homicide rate varied from

about 59 percent of these municipalities experiencing a 1-74 percent decrease to about 9

percent of municipalities experiencing a 75 percent or higher decrease Additionally a modest

number (n=19 7) of municipalities experienced an increase in their homicide rate

30

Exhibit 8 Percent reduction in homicide rate by number of municipalities

Exhibit 9 presents six negative binomial models for the monthly number of homicides The first

model examines the impact of the gang truce and implementation of the Violence Free

Municipality program and our interaction variable that measures the additive effect of both

the gang truce and the Violence Free Municipalities program on the number of homicides We

found that although the gang truce was associated with a significant decline in homicides the

Violence Free Municipality program was related to a significant increase in homicides13

However countrywide the additive effect of implementing the gang truce and the Violence

Free Municipalities program was associated with a significant decline in homicides

13 To be clear our bivariate analyses showed that the violence free municipalities program was related to a significant decline in homicides For example on the one hand those municipalities that did not participate in the violence free municipalities program experienced a decline in their homicide rate from 406 homicides per 100000 population in the pre-truce period to 321 homicides per 100000 in the post-truce period On the other hand those municipalities that participated in the violence free municipalities program experienced a substantially greater decline in their homicide rate from 752 homicides per 100000 population in the pre-truce period to 392 homicides per 100000 in the post-truce period However after other variables are controlled for in our negative binomial models we found that the violence free municipalities program did not have a positive impact on homicides over and above the gang truce itself

31

Exhibit 9 Negative Binomial models for monthly number of homicides

Model 1 Coefficient (se)

Model 2 Coefficient (se)

Model 3 Coefficient (se)

Model 4 Coefficient (se)

Model 5 Coefficient (se)

Model 6 Coefficient (se)

Homicide - 1 month lag 0010 0010 0010 0010 0010 0010

(0002) (0002) (0002) (0002) (0002) (0002)

Homicide - 2 month lag 0012 0013 0013 0013 0013 0013

(0002) (0002) (0002) (0002) (0002) (0002)

Homicide rate spatial lag 0003 0003 0003 0003 0003 0003

(0001) (0001) (0001) (0001) (0001) (0001)

Gang truce implemented -0483 -0541 -0553 0057 -0544 0564

(0066) (0065) (0064) (0278) (0064) (0279)

Violence free municipality 0497 0183 -0003 -0076 -0037 -0094

(0181) (0167) (0152) (0143) (0153) (0145)

Ln( +1) of MS13 prisoners 0066 0176 0062 0171

(0023) (0028) (0043) (0028)

Ln( +1) of 18th St prisoners 0067 0044 0064 -0043

(0021) (0024) (0021) (0024)

MS13 prisoner spatial lag -0004 0054 -0003 0049

(0045) (0055) (0046) (0056)

18th St prisoner spatial lag 0022 0014 0019 0012

(0027) (0032) (0027) (0032)

Gang truce violence free municipality

-0346 -0294 -0309 -0161 -0304 -0059

(0158) (0156) (0158) (0146) (0158) (0145)

Gang truce Ln( +1) of MS13 prisoners

-0185 -0185

32

(0030) (0030)

Gang truce Ln( +1) of 18th St prisoners

0160 0060

(0025) (0025)

Gang truce Ln( +1) of MS13 prisoners spatial lag

-0139 -0138

(0064) (0064)

Gang truce Ln( +1) of 18th St prisoners spatial lag

0023 0023

(0039) (0039)

Ln( +1) of MS13 on street 0023 0018 0021 0016

(0013) (0012) (0013) (0012)

Ln( +1) of 18th St on street 0043 0039 0040 0037

(0016) (0015) (0016) (0015)

Time 0010 0011 0011 0011 0011 0011

(0002) (0002) (0002) (0002) (0002) (0002)

1month 0064 0065 0067 0067 0067 0067

(0055) (0055) (0055) (0055) (0055) (0055)

2month -0020 -0020 -0022 -0023 -0022 -0023

(0056) (0056) (0056) (0056) (0056) (0056)

3month 0088 0088 0188 0088 0088 0088

(0052) (0052) (0052) (0052) (0052) (0052)

4month -0011 -0011 -0011 -0011 -0011 -0011

(0052) (0052) (0052) (0052) (0052) (0052)

5month 0020 0020 0020 0021 0020 0021

(0052) (0052) (0052) (0052) (0052) (0052)

6month 0000 0000 0000 0000 0000 0000

7month 0036 0036 0036 0036 0036 0036

(0052) (0052) (0052) (0052) (0052) (0052)

33

8month 0047 0047 0047 0046 0047 0046

(0051) (0051) (0051) (0051) (0051) (0051)

9month -0055 -0055 -0057 -0058 -0057 -0058

(0053) (0053) (0053) (0053) (0053) (0053)

10month 0040 0040 0041 0041 0041 0041

(0052) (0052) (0052) (0052) (0052) (0052)

11month 0022 0022 0022 0022 0022 0022

(0052) (0052) (0052) (0052) (0052) (0052)

12month -0040 -0040 -0042 -0043 -0042 -0043

(0053) (0053) (0053) (0053) (0053) (0053)

Urban opportunity factor 0209 0042 0033

(0040) (0040) (0039)

male 10-29 0039 0007 0009

(0021) (0019) (0019)

female-headed household -0006 0002 0000

(0007) (0007) (0006)

unemployed 0001 -0002 -0000

(0006) (0005) (0005)

Racialethnic heterogeniety 0058 -0039 -0087

(0263) (0235) (0230)

Intercept -10571

-10544

-11054 -11618 -11015 -11558

(0064) (0060) (0194) (0241) (0212) (0257)

Ln(Alpha) -2287 -2289 -2293 -2294 -2293 -2294

(0113) (0113) (0114) (0114) (0114) (0114)

Truce random effect 0202 0193 0202 0144 0201 0143

(0039) (0038) (0040) (0031) (0040) (0031)

Intercept random effect 0315 0226 0192 0164 0183 0158

(0047) (0035) (0032) (0027) (0031) (0027)

Cov(Truce and Intercept) -0162 -0099 -0122 -0086 -0114 -0090

(0037) (0031) (0031) (0025) (0031) (0026)

34

The second model includes those variables from Model 1 but also includes variables that

controlled for community-level structural factors In this model the gang truce remains

significantly associated with a decline in homicides but the Violence Free Municipality program

and our interaction variable were no longer related to a reduction in homicide Although the

percentage of residents in a municipality who are male aged 10 to 29 female-headed

households percent unemployed and ethnic heterogeneity were unrelated to changes in

homicide the urban opportunity factor was significantly associated with homicide

Models 3 and 4 included our measures of intervention as well as our measures of the number

of MS13 and 18th Street on-the-street and incarcerated gang members at the municipality level

The analysis showed that while the number of MS13 on the street was unrelated to homicides

the number of 18th Street gang members was associated with an increase in homicide These

analyses also indicated that the number of incarcerated MS13 and 18th Street gang members

from a municipality was associated with a significant increase in homicides We further

examined whether this relationship was associated with the gang truce by including two

interaction variables as presented in Model 4 One measured the interaction between the

number of incarcerated MS13 gang members and the gang truce and a second measured the

interaction between the number of incarcerated 18th Street gang members and the gang truce

The results showed that following the gang truce the number of incarcerated MS13 gang

members was related to a significant decrease in homicides and the number of incarcerated

18th Street gang members was related to a significant increase in homicides

Models 5 and 6 in Exhibit 9 present the same two models (3 and 4) as above but include the

community-level structural variables Model 5 once again shows that the number of MS13

members on the street remains unrelated to the change in homicides and the 18th Street

members on the street was associated with a significant increase in homicide However Model

6 shows that following the gang truce the number of incarcerated MS13 gang members

remained significantly associated with a decline in homicides while the number of incarcerated

18th Street gang members remained significantly associated with an increase in homicides

35

Community-level structural factors were found once again to be unrelated to changes in

homicide rates

Exhibit 10 below further illustrates the relationship between reductions in homicides by

municipality and the municipal-level presence of MS13 and 18th Street gang members in

prison The figure shows that in municipalities with no incarcerated MS13 and 18th Street gang

members no change in homicides occurred following the gang truce However when a

municipality had ten MS13 gang members imprisoned on average that municipality

experienced a 55 percent reduction in homicides When a municipality had fifty MS13 gang

members imprisoned those municipalities on average experienced a 71 percent reduction in

homicide following the truce Conversely our analysis showed that the number of 18th Street

gang members in prison (from a municipality) had a significant and positive impact (that is the

number increased) on homicides following the truce For example if a municipality had ten

18th Street gang members imprisoned on average that municipality experienced a 31 percent

increase in homicides14

14 We examined whether there was an interaction effect between gang members on the street following the gangs truce Analysis for both MS13 and 18th street and their interaction with the gang truce showed no effect

36

Exhibit 10 Impact of the gang truce by number of gang members in prison at the municipal level

The Gang Truce and Crime Displacement

We examined two types of displacement crime type displacement and method displacement

Some policymakers have suggested that although homicides may have declined as a

consequence of the gang truce other forms of criminality or categories of reported crime may

have subsequently increased Crime type displacement occurs when offenders who focus on

one type of crime switch to another form of crime in order to avoid detection or to benefit in

some other way (Eck 1993) In El Salvadorrsquos case some critics of the gang truce have suggested

that as homicides decreased other forms of criminality such as extortion might have

increased substantially due to gang membersrsquo increased freedom to conduct activities inside

and outside of prison (Dudley 2013 Parkinson 2014) Method displacement occurs when

offenders change their tactics or methods of conducting crimes as a consequence of an

intervention (Eck 1993) Following the gang truce in El Salvador some analysts argued that gang

37

members might have begun to hide the bodies of homicide victims to avoid detection and to

ldquoprotect the integrity of the countryrsquos gang trucerdquo (Bargent 2013 1)

As noted above we rely on police data that measured monthly numbers of thefts extortions

robberies rapes and auto theftsrobberies by municipality We also received data on the

annual number of disappearances in each municipality by year Exhibit 11 presents the

descriptive statistics for these variables It shows that the rates of theft and robbery did not

change significantly between the pre- and post-true periods and the rate of extortions

significantly declined The t-tests showed that the rates of rape disappearance and

theftrobbery fromof an auto increased significantly

Exhibit 11 Descriptive statistics for measures of displacement

Mean Std Err [95 Conf Interval]

Theft rate

Pre-truce 1238 049 1142 1335

Post-truce 1213 049 1115 1310

Extortions rate

Pre-truce 353 023 308 398

Post-truce 271 021 229 314

Robbery rate

Pre-truce 544 025 495 593

Post-truce 553 025 504 602

Rape rate

Pre-truce 049 003 043 056

Post-truce 064 005 054 073

TheftRobbery Vehicle rate

Pre-truce 132 022 090 175

Post-truce 165 023 120 210

Disappearance rate

Pre-truce 805 068 670 940

Post-truce 1007 064 882 1132

p lt 05

We further examined the data (excepting disappearance data) similarly to the analyses above

in which we used random effects negative binomial models and regressed the number of

38

crimes (ie theft extortion robbery rape and auto theftrobbery) on various months on

predictors with both temporal and spatial lags Our independent and control variables

remained the same as those used in Model 6 (displayed above in Exhibit 9) We found that

over the study period there was no significant change between the pre-truce and post-truce

periods in the number of thefts extortions robberies rapes and auto theftsrobberies (tables

not shown)

As noted above we also received data on the number of disappearances by municipality and

year but because the data were provided by year there were not enough data points to

examine them temporally Therefore we added this covariate (number of disappearances by

municipality and year) to Model 6 in Exhibit 9 The results of the analysis indicated that the

number of disappearances was not significantly related to change in homicides the gang truce

remained associated with a significant reduction in homicides and our interaction variables

(number of incarcerated MS13 amp 18th Street members truce) remained significantly

associated with homicides

Conclusions

The present study sought to examine the impact of the gang truce on violence and other forms

of criminality We examined homicide data by municipality prior to and after the gang truce

Our outcome variables were obtained from the PNC along with several independent and

control variables obtained from the Ministry of Economy and the National Bureau of Prisons

We analyzed this data using a number of time series and random effects negative binomial

models where we regressed the rate of homicide on various months on predictors with both

temporal and spatial lags and controlled for other contextual factors This type of analysis

allowed us to correct for the number of homicides in one month being related to the number of

homicides in the previous month (ie temporal autocorrelation) Additionally it allowed for

the fact that some communities participated in supplemental interventions (ie violence free

municipalities initiative) which might have impacted the outcome in the same way in a given

month These techniques allowed us to isolate the effects of the gang truce as best possible

We also used data from the PNC that measured by municipality the number of thefts

39

extortions robberies rapes auto theftsrobberies and disappearances to examine the

possibility of the trucersquos impact on crime displacement and diffusion of benefits`

Our findings suggest that El Salvadorrsquos gang truce had a significant and dramatic impact on the

homicide rate The mean number of monthly homicides declined from about 354 prior to the

truce to about 218 following the truce for a net decrease of about 136 homicides per month

Our forecast showed that between March 2012 and June 2014 the truce had saved about 5501

lives As will be discussed further below the analysis suggests that key stakeholders have the

capacity to renegotiate existing norms of violence and that at least some gangs have the

capacity to exert substantial informal social control over their members that can result in

reduced violence

The gang truce also lasted substantially longer than previously evaluated truces Ordog et al

(1995) for example reported that the much publicized gang truce in Los Angeles decreased

homicides by about 35 percent for the first three months but then doubled in months four

through eleven compared with the pre-truce period Similar findings were reported in Trinidad

and Tobago (Maguire Katz and Wilson 2013) where it was determined that homicides

declined for a brief period of time (again for about three months) but then increased

substantially over the long term (12 months) These results suggested that gang truces may

produce short-term benefits yet result in long-term adverse consequences

The findings of the present case study suggests that some gang truces might last longer than

previously believed While the number of homicides began to slowly increase about 12 months

following the truce in El Salvador the results clearly showed that almost two years following

the truce homicides still remained below those experienced prior to the truce

We found however that the gang truce did not result in a homogenous decline in violence

across municipalities About 61 percent of municipalities experienced a decline in homicides

but the decline in violence varied substantially between municipalities For example while

about 16 percent of these municipalities experienced a 25 percent reduction in homicides a

number of others (37) experienced a 75 percent or greater reduction Furthermore it is

important to note that about 20 percent of municipalities experienced a modest increase in

40

homicides This suggests that the impact of a gang truce might be variable and could be

dependent on contextual factors We examined the possible influence of these factors by

assessing the impact of social structural factors and the presence of gangs on the municipal-

level impact of the gang truce While we found that social structural factors were unrelated to a

decline in homicides municipal-level gang presence was associated with the decline in violence

as a consequence of the gang truce

We examined this issue further by parsing out the relative influence of the number of MS13

and 18th Street gang members on the street and in prison from each municipality As noted

above we hypothesized that municipalities where gang member presence was high regardless

of their presence on the street or in prison would experience a greater reduction in homicides

because of their increased influence in these areas We found however that the relationship

was not as direct as we would have suspected In particular our analyses indicated that

following the truce the number of MS13 and 18th Street gang members on the street in a

municipality was not significantly related to a decline in homicide but the number of

imprisoned MS13 and 18th Street gang members was associated with a significant change in

homicides following the gang truce

Of special note was our finding of the differential impact of the truce based on gang Although

we found that the number of imprisoned MS13 gang members was associated with a significant

decline in homicides following the gang truce we also found that the number of imprisoned

18th Street members was associated with a significant increase in homicides following the truce

These divergent findings might be associated with each gangrsquos organizational structure and its

capacity to regulate member behavior

Much prior research suggests that imprisoned MS13 gang members have substantial influence

over violence in Salvadoran communities15perhaps even more so than formal mechanisms of

15 Instituto Universitario de Opinioacuten Puacuteblica (Iudop) La situacioacuten de la seguridad y la justicia 2009-2014 entre expectativas de cambio mano dura militar y treguas pandilleras Aguilar Jeannette (Coord) [et al] --1ordf ed --San Salvador El Salv Instituto Universitario de Opinioacuten Puacuteblica (Iudop) 2014- 260 p graacuteficos tablas cuadros y figuras 28 cm (Talleres graacuteficos UCA) httpwwwucaedusviudopwp-contentuploadslibro_la_situaciC3B3n_de_la_seguridadpdf

41

social control such as the police and courts16 Law enforcement officials for years have claimed

that MS13 is one of the most organizationally sophisticated street gangs in the Western

Hemisphere17 The gang has been widely characterized as having a highly vertical organizational

structure and strong control over criminal enterprises in gang-controlled neighborhoods and as

being decidedly capable of enforcing rules through discipline MS13 leadership resides in the

nationrsquos prison system The Ranfla (gang leadership) is comprised of thirteen MS13 gang

members (PNC 2011) who direct coordinate and authorize street crime and other gang activity

from prison Our findings lend support to the idea that MS13 is more organized than the typical

street gang and that imprisoned MS13 gang members exhibit strong influence over their fellow

gang members on the street

Our findings also suggest however that the gang truce had a boomerang effect in

municipalities with high numbers of imprisoned 18th Street members implying that 18th Street

might not have as much organizational capacity to regulate violence on the streets as MS13

The truce provided incarcerated MS13 and 18th Street gang leaders an opportunity to

negotiate with high-ranking officials and influential diplomats This may have increased their

legitimacy inside and outside of their gangs It appears that MS13 was able to exert its span of

control over the communities in which they had influence and they were able to deliver on the

terms of the gang truce negotiations In the case of 18th Street however incarcerated gang

members may not have had the same organizational capacity for communicating and carrying

out directives In fact a review of the gang truce indicated that there was a conflict taking place

between two fractions within 18th Street Consequently the organizational structure and

culture of 18th Street might be more diffuse than that of MS13 and its leadership structure

might not be as strong because of the internal fractures within the gang This might further

explain why homicides increased in 18th Street communities The internal fractures within the

18th Street gang may have resulted in intra-gang violent conflict that was largely contained

16 Instituto Universitario de Opinioacuten Puacuteblica (Iudop) La situacioacuten de la seguridad y la justicia 2009-2014 entre expectativas de cambio mano dura militar y treguas pandilleras Aguilar Jeannette (Coord) [et al] --1ordf ed --San Salvador El Salv Instituto Universitario de Opinioacuten Puacuteblica (Iudop) 2014- 260 p graacuteficos tablas cuadros y figuras 28 cm (Talleres graacuteficos UCA) httpwwwucaedusviudopwp-contentuploadslibro_la_situaciC3B3n_de_la_seguridadpdf 17 httpwwwlaprensagraficacomeua-declara-a-ms-grupo-delictivo-transnacional

42

within 18th Street controlled territories Further analysis is needed to examine this specific

issue

While not the primary focus of this case study we controlled for any impact that the Violence

Free Municipalities program might have had on homicide in 11 municipalities during the truce

As noted above the Violence Free Municipalities program served as the second phase of the

gang truce (Cawley 2013) The program was first proposed by the two mediators and designed

by the Technical Committee on Violence and Crime Reduction which included representatives

from OAS MJPS the mediators and the Humanitarian Foundation (CISPES 2013) The program

involved the mayors of each municipality collaborating with gang leaders to design prevention

and intervention resources for gang members and at-risk youth The Minister of Justice and

Security David Mungia Payes announced that his ministry would facilitate $74 million in

funding from the OAS UN and other donors to implement the programing In exchange gang

leaders agreed to end violence and other criminality in the Violence Free Municipalities (CISPES

2013) The negotiators also agreed to further discuss both gangsrsquo demand to repeal of the

2010 law that increased the capacity of the police and prosecutors to crack down on gangs

(Ayala 2014)

Our findings suggested however that the Violence Free Municipalities program was unrelated

to change in homicides in these communities While much additional research is needed to

understand why the program was not effective it might have been because the gangs had

already agreed to a truce and had already reduced violence to the extent that they could

Conversely the demands made by the gangs as part of the Violence Free Municipalities

program might have been more than could be delivered The time and resources required to

implement the programming and the political capital that was required to repeal legislation

might have been much more than could be delivered Future research is needed to examine the

processes and impact of the Violence Free Municipalities program

Regardless the base rate of violence in El Salvador was reset for a period of time suggesting

that perhaps the gang truce substantially altered existing norms of violence Klein (1995)

argued that cycles of gang violence (that is perceived or real changes in gang activity) are

43

typically the consequence of seasonality epochal variation (peaks and valleys in violence) and

illusion (it appears as if there is a real change in violence but there is not) In El Salvador we

appear to have observed a self-directed cycle of normative change wherein incarcerated MS13

gang leaders directed a reduction in violence by actuating their organizational span of control

through the gang truce Some of the most powerful and influential gang members in El Salvador

used their political social and economic capital to promote the truce and articulate new norms

of violence They were able to do this by leveraging their informal social control over the streets

through actual or perceived threats of violence against those who violated the terms of the

truce

However following a change in government leadership and the government subsequently

distancing itself from the gang truce the conditions and capacity of MS13 leaders to intervene

in local violence might have deteriorated and violence began to increase substantially This is in

part because the third parties in the negotiation were no longer able to communicate with

government officials about furtherance of the truce Specifically the mediators were no longer

able to negotiate on the behalf of the government and were no longer permitted entrance to

the prisons where they could negotiate with gang leaders

44

Case Study B Gang Trucemdash The Jamaican Experience

In the latest report by the UN Office of Drugs and Crime Jamaica ranked as the sixth most

violent nation in the world with a homicide rate of about 521 per 100000 residents (UNDOC

2011) Policymakers and researchers have attributed the nationrsquos high level of violence to such

factors as drug trafficking (Klein Day and Harriott 2004) access to illegal firearms (Lemard and

Hemenway 2006) and historical processes that include a legacy of conflict between the

nationrsquos two primary political parties (Sives 2002 Figueroa and Sives 2003 Moser and Shrader

1999) all of which have facilitated the entrenchment of the more powerful gangs in

communities of the urban poor Likewise high levels of income inequality and chronic youth

unemployment (Francis and Lyare 2006) problematic urbanization (Stone 1975) social

marginalization and an emergent subculture of violence (Harriott 2008) have contributed to the

nationrsquos violence problem Whatever the causes violence has had a considerable impact on

Jamaicarsquos social and economic development it has decreased investor confidence (Schwab and

Porter 2008) tourism (Harriott 2007) and access to public services It has also increased the

costs related to the health care system (Mansinghand and Ramphal 1993) the criminal justice

system (Caribbean Human Development Report 2012) and the education system (Moser and

Holland 1997)

Jamaicarsquos homicide problem is closely associated with its gang problem The Jamaican

Constabulary Force (JCF) has estimated that some 272 gangs are active in the nation most in or

near the capital city of Kingston (Harriott 2014) Gang types and their respective historical

patterns of conflict matter in Jamaica as these variations may determine their predisposition or

amenability to lasting rather than opportunistic truces Jamaican gangs include territorially

organized crime groups conflict gangs defense crews who regard themselves as defenders of

their communities (Levy 2009) and other less cohesive more transient territorial groups Some

of the latter are predatory others bond around the taken identities that generate conflicts of

other similar groups18

18 These are generally rooted in subcultural issues such as the demand to be treated with respect or with the indiscretions and self-centered aggressiveness of individual members that may be associated for example with sexual competition for the favors of women

45

Estimates of the gang-related homicide rate in Jamaica vary perhaps because the crime is not

clearly defined as far as attributing a death to a gang Regardless researchers agree that the

proportion of homicides that are gang related has increased substantially For example Harriott

(2003) reported a fourfold increase in the rate of gang-related homicides between 1983 and

1997 Likewise Hill (2013) using official police data found an eightfold increase from 2001 to

2009 reportedly in 2001 only about 64 percent (n=73) of the nationrsquos 887 homicides were

gang related compared with 523 percent (n=879) of its 1682 homicides in 2009

To address the problem Jamaica has initiated traditional law enforcement strategies such as

establishing a specialized gang unit (Sinclair 2004) initiating curfews (Sinclair and Tuner 2005)

declaring states of emergency (Jamaican Observer 2010) and implementing community-

oriented policing (Kolpack 2006) It also attempted legislative reforms to curb election fraud

and electoral-related violence that involved local gangs (Levy 2009) None of these strategies

stemmed the tide of gang violence In 2002 the Minister of National Security established the

Peace Management Initiative (PMI) (Henry 2011) to augment governmental and non-

governmental organizational capacity to settle gang disputes in the community through

intervention-based programming such as ceasefires and gang truces

The current study examines the peace initiative instituted by the PMI in Greater August Town

Our objective was to understand the negotiation processes undertaken with and between

gangs and other stakeholders Among other things we were interested in identifying the actors

involved in the negotiations the goal(s) of the negotiations and the strategies employed to

carry them out Most importantly however we wanted to determine whether the gang truce

resulted in the desired outcome a reduction in the number of homicides in the Greater August

Town area

The Greater August Town (Jamaica) Peace Initiative

Greater August Town is located on the northeastern outskirts of the city of Kingston The low-

income area has high rates of youth unemployment and a history of gang-related violence

(Charles 2004 Levy 2009) Over the last decade the Greater August Town community has

sought improved living conditions and revitalization (Levy 2009 95) The arearsquos inherent

46

resilience has been augmented by nearby intellectual and cultural engines such as the

University of West Indies the University of Technology and University Hospital (Charles 2004

38)

Greater August Town is comprised of the communities of August Town proper (which is

fractured into several locales with gang-given names such as ldquoVietnamrdquo and ldquoOpen Landrdquo and

city government-approved names such as Hermitage Goldsmith Villa (Angola) and Bedward

Gardens These socially defined community divisions and subdivisions are markers for the

territorial boundaries of street gangs and therefore in some instances are lines of potential

conflict Some of those boundaries demarcate areas of Greater August Town that are

predominantly supportive of one or another political party but the boundaries do not always

hold political significance politics is but one element in the conflict geography of the area Like

many communities of the urban poor the Greater August Town area is easily mobilized

politically ndash a reality that is understood and at times exploited by street gangs who politicize

gang ldquowarsrdquo in their efforts to build alliances and to neutralize the police In fact the basic

principle of community mobilization in Jamaica is political patronage and clientelism Access to

resources (eg jobs housing education) for the poorest residents is often determined by the

local political party Thus according to Charles (2004 36)

supporters attach themselves to the political parties to get first preference in the

distribution of scarce resources and over time because they are unemployable they

become dependent on their political party for their economic survival These supporters

will kill anyone who threatens the support base of their political party because they

perceive it as a threat to their daily survival

As a consequence some political supporters invest heavily in the electoral contests and

provoke conflicts that affirm their loyalty to their party in order to secure material benefits

from it Political competition is one conflict fault line in what otherwise is a politically

heterogeneous community Specifically political support in Greater August Town is divided

between the Peoplesrsquo National Party (PNP) and the Jamaica Labor Party (JLP) The PNP receives

strong support from residents in August Town proper the upper region of Goldsmith Villa and

47

Bedward Gardens while the JLP is supported by those living in Hermitage and the lower region

of Goldsmith Villa (Charles 2004 ) The division between the political parties in the area as in

much of Jamaica was believed to be at the root of much of the communityrsquos violence

particularly between 1979 and 1993 Some gangs aligned with the PNP and others with the JLP

this often resulted in political boundaries overlapping with gang turf (Charles 2004)

Pre-truce Violence

Nationally the history of gang violence in the Greater August Town area first appeared as

political violence closely associated with the electoral cycle As in many other urban

communities the problem intensified mdash deeply affecting community life in the period just prior

to the national elections of 1980 mdash then continued cyclically until 1993 Gangs have since

harnessed this legacy in order to establish and maintain community support based on common

political affiliations They have used that support to nullify the efforts of law enforcement to

suppress their illicit activities (Harriott 2008)

The conflict profile of these gangs and of the communities in which they are nested has

changed over the decades From the beginning of their involvement in political violence and

territorial control a form of gerrymandering existed that manipulated the voting population

forcing certain individuals out of a particular constituency and preventing those who remained

in the community from voting for the opposing party This was linked both to the electoral cycle

and to the types of inter-gang conflicts that typically were associated with street gangs

(Figuerou Harriott and Satchell) More recently however much of the crime and violence

perpetrated by gangs has resulted from internal conflicts (eg status management disputes

over womengirls or money individual membersrsquo activities that could attract police pressure)

Internal conflict at times has led to gang fragmentation and new alliances that pull more parties

into the conflict escalating the homicide rate and increasing the sense of insecurity among the

general population (Levy 2012)

The most significant of these internal conflicts resulted from the killing of former Jungle 12

leader Neil Wright by members of his own gang Jungle 12 was the most influential gang in

Greater August Town Before his murder in order to increase the gangrsquos access to illicit

48

opportunities in Kingston Wright had been attempting to extend Jungle 12rsquos influence with a

system of alliances with other gangs and recruitment in Kingston (Harriott 2014) In short his

ambition was to transform Jungle 12 from a neighborhood street gang operating on the

outskirts of Kingston into to a dominant organized crime network that could reach into the

heart of the city In pursuit of this goal Wright recruited members from outside August Town

elevating them in the gang hierarchy above the locals This led to status-related conflicts and

resistance to Wrightrsquos leadership within the gang His murder precipitated a split of Jungle 12

into three factions two of them fled to other neighborhoods within August Town (Angola and

Vietnam) resulting in the formation of new alliances and a new conflict geography that

replaced the former political geography of conflict Wrightrsquos killing and the subsequent demise

of Jungle 12 as the dominant gang in Greater August Town altered the balance of power and

escalated inter-gang violence (Harriott 2014) The post-2005 phase of conflict was

characterized by power symmetry conflict intensification and the spread of conflict

throughout the entire geographic area of Greater August Town

Although their origins are unclear retaliatory killings and other violent incidents progressively

intensified between 2005 and 2008 The violence was episodic retaliations were most often

motivated by suspicions related to personal and geographic connections between warring

gangs As the violence escalated new alliances were formed to enhance power and dominance

which in turn increased the number of gangs and gang members involved in the violence

(Harriott 2014) This eventually attracted national attention and triggered community

mobilization for a gang truce

The Truce-making Process

The Greater August Town gang truce was preceded by frequent intense violence and public

outrage As noted above the violence had escalated in November 2005 when Jungle 12 leader

Neil Wright was killed The defection of a Jungle 12 member to Goldsmith Villa (Angola) caused

infighting within the gang and conflict between it and Goldsmith Villa Just a few months later

Wrightrsquos brother Steve and two others were injured during a turf battle (Martin-Wilkens 2006)

49

Thereafter violence began to occur at regular intervals until January 2007 when the Peace

Management Initiative (PMI) hosted a peace march in the community Two PNP politicians

urged the community to unite A PMI leader declared that the peace march was being held to

ldquodemonstrate to the public that Jungle 12 members are back together and that they want

peacerdquo (Thompson 2007 1)

Although hopeful some residents remained skeptical about the peace march perceiving the

action to be politically motivated In the absence of trustworthy information inter-group

conflicts tended to be interpreted through a politically partisan lens this created obstacles to

isolating the gangs building a consensus for peace and unifying community mobilization The

politically based narratives weakened the communityrsquos leverage for peace as well as the

exposure of the gangs to police action As one resident said ldquoThe election is coming up and

they want[ed] the people to vote for the PNP is one of the main reasons why they have to walk

todayrdquo (Thompson 2007 1) Those who shared such views stayed away from the peace march

Although that widely held myth was not factual it did serve to demoralize and demobilize one

part of the community A local UWI faculty member was articulate in his assessment of the

politics behind the march

August Town violence is not violence of organized crime which is based on drug

trafficking extortion or some other criminal enterprise [but] rather the violence in

August Town is essentially ldquotribalrdquo mdash the Peoplersquos National Party tribe versus the

Jamaica Labour Party tribe [which has been complicated by a splintering within the

PNP tribe] (2007 1)

Indeed it was not the violence of organized crime but neither was it political violence The

individual quoted above neglected to mention that the conflict was between those pro-PNP

splinter groups who were largely comprised of members of Jungle 12 Moreover their ldquopro-

PNP-nessrdquo was unrelated to the conflict there was no factional infighting within the local PNP

organization at that time

Nevertheless following the peace march the gang violence diminished Then in November

2007 a turf war erupted between two gangs from the Greater August Town neighborhoods of

50

Vietnam and River This time as the police stated the gang violence was less about politics and

more about dominance and turf Police were dispatched to perform directed patrols but

whenever they were not present the shootings continued (Mcleod 2007) In April 2008 the

community witnessed local gangs engaging in a five-hour-long street battle that left two killed

and three others wounded It ended only after the police deployed armored trucks The next

month another round of gang violence resulted in five others being killed including a one-year-

old child This resulted in the three members associated with the gang who committed the

homicides being killed in retaliation (Virtue 2008) The local community mobilized against the

violence increasingly cooperating with the police providing more information about the gangs

Subsequently gang members observed a decline in their influence within the community

During the early period characterized by low-intensity conflict the less influential gangs at

times used manipulation of the police as a tactic for suppressing the more influential gangs

This was largely done through strategic release of information Prior to 2005 when Jungle 12

was dominant its membersrsquo illicit activities were constantly reported to police by members of

other gangs as a means of compelling a compromise or settlement of conflicts In practice this

was done by ldquotrading casesrdquo Once a crime had been investigated by the police and suspects

had been charged an opportunity was created for the gangs and other parties to the conflict to

settle the matter by agreeing to drop their cases (typically by ceasing cooperation with police

investigators) This type of ldquoself-helprdquo served to end some of the retaliations but it rested upon

the manipulation of the police (Harriott 2014)

Later in an attempt to quell escalating inter-gang violence the police established buffer zones

between the warring gangs This action resulted in unintended consequences For example

when the police declared a buffer zone between August Town and Hermitage Hermitage took

advantage of the opportunity to attack Angola Some Angola residents accused the police of

turning a blind eye and creating an opportunity for Hermitage to attack their community

Although little reliable information exists about why the police made the deployment the way

that they did it is more likely that the police inadequately assessed the situation (ie mis-

assessed the pattern of alliances and the likely targets of attack) (Harriott 2014) In the areas

affected by this kind of increasing violence community members became angered and lost

51

confidence in the police The error resulted in some parties to the conflict receiving increased

support from their communities and in greater gang-community cohesion (Harriott 2014)

After a brief period the police identified this problem and began to disengage by no longer

providing a buffer between gang controlled areas which in turn allowed still more conflict to

occur between the gangs

As the violence escalated beyond their control police finally responded by applying their own

forms of pressure For example units under the direction of the JCF High Command would

make periodic raids in the community during which they would at times seize weapons and

make mass arrests (Sinclair 2005) However there were also moments when the local police

were very responsive improved their relations with the community and consequently gained

greater access to relevant information Two such moments occurred just prior to and again

immediately after the truce moments during which there was greater freedom of movement

and open collaboration between the community and the police (Harriott 2014)19

The Establishment of the Greater August Town Gang Truce

The Greater August Town gang truce was led by the Jamaican Peace Management Initiative

(PMI) The PMI is a government-funded initiative created for the purpose of working with gang

members to reduce violence Due to community mistrust of the police in 2002 the organization

was established as an alternative organizational mechanism for responding to gang violence

The PMI sought to bridge government and civil society efforts to mediate disputes between

gangs as well as to provide outreach to gang members (Bakrania 2013) While efforts to

institute a gang truce in Greater August Town were led by the PMI a number of other

stakeholders helped to facilitate the truce these included faculty at the University of the West

Indies (UWI) and representatives from the police the faith-based community and the August

Town Sports and Community Development Foundation (Jackson 2008 Levy 2009 also see

Appendix C) The gangs involved in the truce included those from August Town Hermitage

19 The quality of police-community relations largely depended on the style of the local station commander however regardless of the external environment

52

Goldsmith Villa Bedward Gardens and African Gardens Because of its formality as well as its

perceived effectiveness the truce signed on June 24 2008 was regarded by many as the first

of its kind in Jamaica (Levy 2009)

Truce negotiations began early in June 2008 and lasted for about three weeks The gangs

sought to leverage their violence-making capabilities and demanded payment for peace They

asked the third-party negotiators for money ldquoworkrdquo and start-up funds for proposed micro-

businesses (Wilson 2014) Those demands were rejected by the negotiators on the grounds that

the third-party institutions would not buy a peace that was intended to save the lives of those

who were making the demands Moreover if peace was to be purchased then gang conflict

could be used continuously to extract money and other benefits from negotiators The third-

party actors made some demands of their own In some quarters of the community and society

the surrender of guns was viewed as a litmus test of the sincerity of the gangs Consistent with

this the negotiators suggested that the parties to the conflict symbolically hand over one gun

each that suggestion was immediately rejected by the gang leaders These kinds of demands

from the various parties ceased after a time as they all agreed that the truce was to stand on

its own merits (Harriott 2014)

As the truce began to be committed to paper a number of stakeholders expressed concern that

their greatest risk in participating could be the potential for Jungle 12 factions to use the peace

agreement as they had in the past as a tactic to persuade their enemies to let their guard

down Others however recognized that Jungle 12 had now been weakened and that a formal

public peace agreement would be beneficial to the gang and therefore this time would be

different (Harriott 2014)

The gangs held fast to their claim that their weapons were needed for their own protection

because the police were ineffective in responding to violence in their communities (Jamaican

Gleaner 2014) It became a precondition of the truce that the gangs would not be required to

turn in their guns and other weapons (Jackson 2008) The truce agreement did specify

however that ldquoall persons are allowed to move freely across all boundaries regardless of

reputation or affiliation No gun salute or any other shooting is to take place in the community

53

for a period of at least five yearsrdquo (2008 also see Appendix C) The truce agreement and its

conditions were prescribed in a document that was finally signed by all of the major

stakeholders including the gangs (see Appendix C)

Throughout the negotiations each of the gang leaders had attempted numerous times to use

the truce as an opportunity to bargain for money jobs and business support grants Such

demands consistently were rejected by the third-party actors Nonetheless both prior to and

after the truce some efforts were made to create better opportunities for young people

residing in the community UWI for example provided a homework supervision program to

encourage students to further their education and it developed a community-building initiative

to help improve schools and enhance sports programming (Levy 2009) Such programs were

conducted as part of UWIrsquos Township Project in August Town which invested significant

resources in developing the residentsrsquo job-related capabilities and collective self-efficacy

The Greater August Town gang truce was noteworthy for two reasons First the gang truce

received substantial press attention The media were invited to witness the ldquosigningrdquo of the

truce by the gang leaders in the presence of a JCF Deputy Commissioner of Police a PMI board

member and the UWI Principal and two professors of its faculty Second the truce was widely

credited with decreasing violence in Greater August Town and it served as an exemplar to

other communities seeking to replicate its success (Virtue 2008) A number of reports

manuscripts and newspaper articles proclaimed the truce to be a success Bakrania (2013 10)

for example reported that ldquoPMI has been credited with stopping gang wars in August Town

rdquo Levy (2009 94) remarked that the ldquomost interesting outcome of PMI efforts to date was the

Peace Agreement reached in August Town in late 2008rdquo Likewise a government report noted

that ldquothe peace treaty was a pivotal achievement in August Town that has significant potential

for wider application Crime levels dropped markedly in August Town after the signing of the

peace agreement in June 2008rdquo (McLean and Blake-Lobban 2009 78) To this day August Town

celebrates the signing of the truce with an annual celebration with food and music

(Cunningham 2011)

54

Methods

Our evaluation relied on a pre-testpost-test quasi-experimental group design Our

methodology examines the Greater August Town community which is comprised of three

contiguous towns where the gang truce took place (the target area) and the balance of Jamaica

which is comprised of 178 communities (comparison areas) As seen in Exhibit 12 the average

number of residents living in each of the three communities in the target area was not

significantly different than that for the rest of Jamaica about 7776 residents lived in each of

the Greater August Town communities compared with 6468 in the other communities

Likewise communities of Greater August Town were about as densely populated as other

communities (2960 per square kilometer versus 2647 per square kilometer) and the age

range of residents was similar as well However Greater August Town (a) had a significantly

higher proportion of its residents living in poverty (196 vs 158) (b) consumed fewer

resources than other communities and (c) reported significantly more homicides than other

communities prior to the truce (see Exhibit 12)

55

Exhibit 12 Descriptive characteristics of Greater August Town and balance of Jamaica (2007-2011)

Comparison

Area Greater August

Town All areas

Population (mean) 6468 777633 648994

(sd) 720482 353731 715621

Population density (mean) 264719 296033 265238

(sd) 271023 285501 270465

Percent in poverty 1577 1957 1583

(sd) 1036 106 1029

Consumption 15737890 1106939 1566048

(sd) 10713020 205336 1064021

residents under 15 yrs old 2369 2494 2371

(sd) 487 115 484

residents 15-65 yrs old 6840 6901 6841

(sd) 423 29 419

Murder per month (mean) 674 857 677

sd 1928 1409 1920

Total murders 10068 180 10248

n 178 3 181

plt=05

Measures

Two distinct data sets were merged to measure the impact of the Greater August Town truce

First data from the 2011 decennial census provided community-level measures of the social

and economic characteristics of the 181 communities in Jamaica Described in detail below the

community-level data used in the study included population population density gender age

poverty and consumption20 These data were obtained directly from the Statistical Institute of

Jamaica

Second police homicide data from the years 2007 through 2011 were used to construct the

studyrsquos community-level measure of homicide The homicide data were aggregated by month

20 Consumption is an alternative measure of poverty in Jamaica which measures the consumption of food and non-food items

56

and appended to the community-level data The final (merged) data set included 10248

homicides over the 60-month study period These data were obtained from the Jamaica

Constabulary Force (JCF)

The dependent variable examined in the study was constructed from official police homicide

data Once again the homicide data represented the number of officially recognized homicides

in Greater August Town and each of the remaining communities in Jamaica We examined

change by comparing the homicide data 18 months prior to the truce with the homicide data 42

months following the truce More specifically we examined whether there was a change in the

number of homicides in the 30 days following the truce (month 1) as well as whether the truce

had an impact every three months thereafter (ie months 2-5 6-8 9-11 12-14 15-42) and

whether any changes in homicide coincided with changes in homicide in the balance of

observation areas The frequency distribution of our dependent variable is presented in Exhibit

13 It shows that prior to the truce the target area on average experienced significantly more

homicides (1495) than did the comparison areas (920)

57

Exhibit 13 Distribution of homicides in the target and comparison areas

Comparison

Area Target

Area Total

Pre-truce period Mean 920 1495 932

SD 2469 1966 2461

N 241400 5100 246500

Month 1 of truce Mean 741 286 733

SD 1785 496 1772

N 17800 300 18100

Months 2 thru 5 of truce Mean 647 905 652

SD 1821 1249 1812

N 71200 1200 72400

Months 6 thru 8 of truce Mean 577 1236 588

SD 1690 1074 1683

N 53400 900 54300

Months 9 through 11 of truce Mean 718 333 711

SD 2034 733 2019

N 53400 900 54300

Months 12 through 14 of the truce Mean 687 095 678

SD 1519 286 1509

N 53400 900 54300

Months 15 thru 42 of the truce Mean 564 589 564

SD 1683 1042 1674

N 516200 8700 524900

Total Mean 674 857 677

SD 1928 1409 1920

N 1006800 18000 1024800

An illustration of the trends in homicide prior to and following the gang truce are shown in

Exhibit 14 It shows that 30 days following the truce homicides fell in the target and

comparison areas then increased and decreased several times with a general downward slope

in violence over time

58

Exhibit 14 Monthly number of homicides pre-post truce in the target and comparison areas

We also used a number of measures to control for community-level structure from the 2011

decennial census These community-level data included the communityrsquos population

population density (per square kilometer) and community level of consumption Additionally

the census data included measures of the percentage of the population that was female under

15 years old 15 and 65 years old and 65 years old and older as well as a measure of the

percentage of the population living in poverty Principal components analysis was used to

reduce some of these data into a summary measure

Exhibit 15 shows the results of the component loadings One component was extracted that we

designated as socio-economic status (SES) which exhibited high loadings for percent living in

poverty percent under 15 years old percent 15 to 65 years old and consumption Excluded

from the principal components analysis were population and population density Population

was used as our exposure variable and population density was logged to address skewness in

these data

Exhibit 15 Factor loadings from principal components factor analysis

Loading Poverty 78 Consumption -76 under 15 years old 92 between 15 and 65 years old -80

59

Analytic Strategy

In order to test whether the truce had an impact on homicides in the target area andor

whether displacement had occurred in the balance of the study area several analytic

techniques were employed Most of the methods employed the use of the homicide rate as the

dependent variable We explored the data in this way to provide the maximum statistical

power to detect an effect As a check on these methods we also employed a generalized model

to compensate for the non-normality of our outcomes

First focusing only in the target area we performed a simple t-test comparing the homicide

rates before and after the truce (the unit of analysis was a month) however this technique had

limitations the most severe of which was that even if the test were significant it would be

difficult to determine whether the difference was due to the gang truce or a natural change

over time in the outcome Second to address this limitation time series models were

employed whereby the homicide rate for the target area was modeled as a function of time

with truce period indicators included to measure the effect of the truce net of the temporal

trends These models were estimated with ARIMA techniques with a one-month lag auto-

correlated error Third we examined the homicide rate for each town using a panel time series

model In this model the temporal trend for each town was examined with indicators for target

areas and truce periods included The main effects for the truce periods measured the effect of

the truce in the target areas and the moderators of the truce period in the comparison areas

measured displacement effects Finally because the dependent variable coded is not normally

distributed across months we used a negative binomial time series model to estimate the

number of homicides with the population covariate serving as an exposure variable

Findings

The first set of results examines only the target area The first test was a simple t-test

comparing the mean homicide rates before and after the truce periods The result was a mean

difference in the homicide rate of -890 per 100000 with a significant t-statistic of 370 While

60

this result is statistically significant we caution that it may or may not reflect an impact of the

truce To further examine the truce effect in the target area we performed ARIMA regressions

The first model did not include an effect of the temporal trend In this model we again find that

by month 15 the murder rate decreased by about -89 per 100000 (Exhibit 16)

Exhibit 16 Results of basic ARIMA model

Next we employed the ARIMA model again but included a variable (date) to control for the

temporal trends in the data Exhibit 17 shows that when we controlled for temporal trends the

impact of the truce we observed was no longer significant This result indicates that it was not

the truce per se that caused the decline in homicides but instead the decline in homicides was

part of a larger (local and nationwide) decline in homicides

Note The test of the variance against zero is one sided and the two-sided sigma 7877339 5907266 1334 0000 6719536 9035142 L1 1720758 1242218 139 0166 -0713944 415546 ar ARMA _cons 1538436 1904285 808 0000 1165203 1911669 t_15 -8888017 2969383 -299 0003 -147079 -3068134 t_12 -1507749 1820406 -083 0408 -507568 2060181 t_9 -132475 9189969 -144 0149 -3125951 4764507 t_6 -2861008 6335492 -045 0652 -1527835 9556328 t_2 -5166363 4498142 -115 0251 -1398256 3649834 t_1 -1111771 4463623 -025 0803 -9860311 7636769murders_rate murders_rate Coef Std Err z Pgt|z| [95 Conf Interval] OPG

Log likelihood = -2089969 Prob gt chi2 = 00397 Wald chi2(7) = 1472Sample 564 - 623 Number of obs = 60

ARIMA regression

61

Exhibit 17 Results of ARIMA model with control of temporal trends

We next estimated the possible displacement effects of the truce Exhibit 18 presents the

results of these models Examination of the main effects of the truce period does not indicate

any effects and looking at the truceComparison interaction effects we also do not find any

displacement effects Note that these models also controlled for the socio-demographic

characteristics of each community

Note The test of the variance against zero is one sided and the two-sided sigma 7775398 6622814 1174 0000 647735 9073446 L1 1270618 1253989 101 0311 -1187156 3728392 ar ARMA _cons 1480274 1197324 124 0216 -8664378 3826986 date -2318963 2091772 -111 0268 -6418761 1780835 t_15 -3519687 7694691 -005 0964 -1543329 1472935 t_12 -970747 2228007 -044 0663 -5337561 3396067 t_9 -8844212 9402792 -094 0347 -2727335 9584921 t_6 2995697 6576644 005 0964 -1259042 1318956 t_2 -2353533 4923798 -048 0633 -12004 7296933 t_1 -9194625 5462116 -017 0866 -1162501 9786088murders_rate murders_rate Coef Std Err z Pgt|z| [95 Conf Interval] OPG

Log likelihood = -2082031 Prob gt chi2 = 00710 Wald chi2(8) = 1444Sample 564 - 623 Number of obs = 60

ARIMA regression

62

Exhibit 18 Results of Panel (Town) time series model with control of temporal trends

Last we used a random effects negative binomial regression that predicted the homicide rate

with population as an exposure variable and controlled for the socio-demographic

characteristics of each community The results are presented in Exhibit 19 The analysis showed

that time had a negative effect indicating that homicides were decreasing in all areas over the

study period The main effects of the truce (truce = 1 2 ) represented the effects of the

truce in the targeted area and did not show a significant effect for any period following the

gang truce However we did find that the homicide rate significantly increased in the

_cons 6310151 3518738 179 0073 -5864483 1320675lnpopulationdensity 121961 1709831 713 0000 8844891 1554731 ses 1245334 2428629 513 0000 7693312 1721336 date -1276623 0345659 -369 0000 -1954101 -0599144 15Balance 5392222 4128863 131 0192 -2700202 1348464 12Balance 1318516 7903159 167 0095 -2304746 2867507 9Balance 1042658 7925876 132 0188 -5107854 2596101 6Balance -1703928 7919006 -022 0830 -1722489 1381704 2Balance 3498997 7175155 049 0626 -1056405 1756204 1Balance 1056874 1104696 096 0339 -1108291 3222039 truceaugust Balance -4757021 3276618 -145 0147 -1117907 1665032 august 15 -4269255 4284086 -100 0319 -1266591 41274 12 -1262496 7871907 -160 0109 -2805361 2803697 9 -1014276 7881965 -129 0198 -2559113 5305603 6 6563019 786731 008 0934 -1476334 1607595 2 -5034368 7123478 -071 0480 -1899613 8927392 1 -1215158 1095699 -111 0267 -3362688 9323715 truce murders_rate Coef Std Err z Pgt|z| [95 Conf Interval]

Prob gt chi2 = 00000 Wald chi2(16) = 12359 max = 60 avg = 5661878Estimated coefficients = 17 Obs per group min = 48Estimated autocorrelations = 1 Number of groups = 181Estimated covariances = 1 Number of obs = 10248

Correlation common AR(1) coefficient for all panels (02045)Panels homoskedasticCoefficients generalized least squares

Cross-sectional time-series FGLS regression

63

comparison areas in months 12 through 14 following the truce In particular we found a 29

percent increase in the homicide rate in the comparison communities for that period (exp (-

1797 + 2048) = 1285 p-value = 004) Since this effect is only significant at the 005 level

however and given the number of analyses used to examine the data it is possible that we

found this effect by chance alone

64

Exhibit 19 Random Effects Negative Binomial

65

Given these caveats we visualized this model with the following set of marginal predictions as

observed in Exhibit 20 We saw that the targeted area (as illustrated in red) experienced an

immediate decrease in homicide which coincided with an increase in homicides in the balance

of the study area However the target area quickly returned to ldquonormalrdquo and homicides in the

comparison area decreased again During months 9 through 11 following the truce there was a

reduction in homicides in the target area with an associated increase in the comparison area It

is important to point out that the confidence intervals are large and we cannot yield concrete

conclusions from these results However it appears that the truce might have had a temporary

short-lived displacement effect decreasing homicides in the target area but increasing

homicides in the comparison area

Exhibit 20 Predicted change in homicides in the target and comparison areas

Conclusions

From 2000 through 2009 Jamaica experienced a substantial number of homicides many of

which were attributed to gangs in one form or another Traditional law enforcement responses

were repeatedly implemented but until 2010 those had little effect Some policymakers in

Jamaica as well as in other nations throughout the Caribbean and Central America have

recently been experimenting with novel approaches to reducing gang-related violence notably

the implementation of gang truces In Jamaica at least eight gang truces reportedly have been

66

negotiated since 2001 (Levy 2009) The Greater August Town gang truce was thought to have

been one of the more successful and it has served as a model for other communities to use

(2009) Our purpose here has been to identify the actors involved in the negotiations of that

truce the negotiation goals and the implementation methods used and then to examine

empirically the impact of that truce on homicide rates in the targeted community

The 2008 gang truce in August Town was a response to violence that arose when the leader of

one gang was killed creating a power vacuum that other gangs saw as an opportunity to

increase their influence in the community Concomitantly the community as well as the gangs

feared that an absence of formal social control would result in further violence The police

reacted unevenly At some times they engaged in appropriate but heightened levels of

preventive patrol while at other times they purposely provided little or no protection on

occasion they used aggressive tactics that further isolated them from the community The end

result was that there was neither stability nor predictability in the police response and

therefore little trust in the police to address the problem

As the violence further escalated the community mobilized The Jamaican Peace Management

Initiative faculty members from the University of the West Indies (UWI) the Jamaican

Constabulary Force (JFC) and other community-oriented groups joined forces seeking to

reduce the increasing number of homicides by brokering a truce between the gangs Over the

three-week negotiation period the negotiators and the gangs sought terms from one another

The gangs wanted payment ldquoworkrdquo and funds for micro-business development to end the

violence The third-party stakeholders wanted the gangs to disarm actually or symbolically

Neither the gangs nor the stakeholders had substantial leverage nor did they have much to

offer one another in terms of incentives In the end however a truce was agreed upon and all

of the gangsrsquo leaders and several key community stakeholders signed it at a public ceremony

with the media in attendance

At first glance our impact findings appeared to show that the gang truce was an effective

mechanism for reducing violence Bivariate analyses showed a significant decline in homicides

after the truce was implemented This explained the work previously published by

67

policymakers researchers and news reporters Upon further examination of the data however

comparing change in the target and comparison areas and accounting for temporal trends we

found that the decline in homicide was part of a larger nationwide decline in violence and that

the gang truce was not responsible for the decline The only significant effect that we

uncovered was that possibly the homicides were displaced outside the target area for a brief

period of time but then returned to normal

Any one of a number of explanations might be offered for the strategyrsquos lack of effectiveness It

might be that the Jamaican gang leaders at least those in Greater August Town did not have

the organizational capacity to change gang member behavior Much prior research suggests

that in general gangs have limited organizational structure and little formal leadership This

might suggest that gangs do not possess the necessary capacity to regulate their membersrsquo

violence That said gangs in Jamaica including in Greater August Town have been found to be

fairly organizationally sophisticated and to possess strong leadership

In fact in a small number of Jamaican communities gangs have been found to be highly

organized with individual gang leaders being referred to as dons and community leaders The

gang leader in such a community is often found to have substantial control over members and

residents as these communities often turn to the don rather than the police for justice The

don will hold court and punish those who commit crime Punishment can include beatings and

torture as well as execution (Morgensen 2004) Although this level of organizational structure

and sophistication is found only in a small number of Jamaican communities generally the

gangs in Jamaica are believed to have some organizational capacity or at least enough to

reduce violence in communities

Our findings however indicated that prior to 2005 and the death of Neil Wright perhaps only

Jungle 12 could approximate that capacity to discipline members and enforce a truce After the

gangrsquos fragmentation in 2005 Jungle 12 lost much of its organizational capability and

enforcement of the truce was therefore difficult The truce negotiators sought to address the

enforcement issue by proposing a peace council that would involve all parties The proposal

was approved by all key stakeholders still some gang leaders demanded cash payments as a

68

condition for attending council meetings Peace was consistently seen by then as a bargaining

tool rather than as an honest attempt to establish and maintain peace In the end members of

only two gangs were attending the meetings21 and the council soon dissolved

In an effort to replicate the council function UWI sponsored one of the most respected

negotiators a community activist to become a one-person monitoring and intervention

specialist or a ldquoviolence interrupterrdquo His job was to ensure that truce violations did not lead to

a return of the gang wars mdash and there were many violations of the truce For example there

were instances of gang members crossing boundaries and entering the turf of another gang

armed although not initiating conflict behavior that was interpreted by the opposing gangs as

preparation for the next round of ldquowarrdquo or as laying a foundation for a surprise attack that

would exploit the truce for this purpose In the absence of the council these matters were

reported to the violence interrupter who tried to resolve the problems in consultation with the

various gang leaders Often the gang leaders were unresponsive or incapable and therefore the

threatening practices and violence continued Ultimately there were no rules or bodies or

persons who could regulate the violence and there were never any reference points for

compliance The formal truce agreement was an attempt to negotiate and impose such rules

via a collective pressure that would include third parties but it was unsuccessful in doing so

The potential for re-engineering norms related to conflict thus was not realized

Another explanation for the failure of the gang truce might be that it was more a vehicle for

rhetoric rather than for reality The gang leaders insisted that they would sign the truce

agreement only if it were ratified in public with the presence of the media (Jackson 2008 Levy

2009) The leaders might have viewed the process in and of itself as a means of increasing their

reputation and influence within the community and in policymaking circles (and to reduce

mutual distrust) In signing the truce gang leaders publicly pledged to reduce their involvement

in violence thereby calming local residentsrsquo fears They also made public efforts to increase

resources for their communities perhaps in an attempt to portray themselves as ldquoprovidersrdquo to

the community In fact the truce did provide gang leaders with an opportunity to be seen in

21 Interestingly the Jungle 12 factions did not attend any of the peace council meetings

69

public collaborating with important community stakeholders The imagery of the public signing

was of the government (via the PMI) and others approaching the gang to ask them to use their

means of informal social control in the community to reduce violence mdash to accomplish

something that the government could not do on its own As a consequence the process may

have been perceived by gang leaders as a victory because it enhanced the gangsrsquo reputation

with both the government and community

Alternatively from the start the gangs might not have been fully invested in the gang truce

One of the major criticisms of the Greater August Town gang truce was that gangs were not

required to give up their firearms although some believed that this was an unrealistic request

their demand and the demand of many that all guns be turned in immediately was

quite unrealistic given the decades of ingrained gun culture and the continued inability

of the security forces to guarantee protection for any corner against armed rivals It was

obvious to most observers that that kind of situation could not be ended overnight and

that this was a reasonable first step in the process (Levy 2009 63)

The gangs feared that if they were to disarm themselves they would be vulnerable to other

gangs and unable to protect themselves a concern that appears not to have been addressed by

mediators Indeed at times some elements within the community felt somewhat dependent on

the gangs to maintain security If the gangs would have been disarmed and there were no

near-term alternative prospects for any form of social control both the gang and the

community might have faced additional violence as has been observed in the past In the end

the gang truce only called for a reduction in gang violence and did not provide any solutions to

address the larger problems between the gangs nor did it provide the gangs with any tangible

benefits for abiding by the truce

70

Case Study C Gang Trucemdash The Honduran Experience

Introduction

Violence in Honduras is at epidemic levels increasing almost 44 percent over the past five

years In 2012 there were 7172 homicides in Honduras or about 86 homicides per 100000

population (Instituto Universitario de Democracia Paz y Seguridad 2013) making it the most

violent nation in the world (United Nations 2013) Likewise Hondurasrsquo second largest city San

Pedro Sula has the highest municipal level homicide rate in the world with 1290 homicides

(Instituto Universitario de Democracia Paz y Seguridad 2013) or about 174 per 100000

population (United Nations 2013) In comparison the average homicide rate across the globe is

about 62 per 100000 and the average homicide rate in Central America is about 27 per

100000 (United Nations 2013)22

Much of the discussion about the causes of Hondurasrsquo high homicide rate has focused on its

relationship with international drug trafficking routes gangs and conflict between crime

groups and the government and government instability Estimates of gang involvement very

widely but some have suggested that there are between 12000 (Seelke 2012) to 36000

(Ratcliffe et al 2014) gang memberrsquos in Honduras who typically belong to one of two gangs

MS-13 and 18th Street These gangs are said to be less organized than their counter parts in El

Salvador but are said to be just as involved in extortion and intimidation and perhaps more

involved in drug trafficking because of their stronger linkages with Mexican drug cartels

(Wilkinson 2013)

22 An unusual characteristic of the homicide problem in Honduras is the age of victims Typically in the Western Hemisphere homicide victims are aged 15 to 29 In Honduras however those 30 to 44 have the highest rate of violent victimization For example 1 out of 280 males 30 to 44 years old are the victim of homicide compared to 1 out of 360 males 15 to 29 years old (United Nations 2013) These findings by themselves are suggestive of a chronic gang problem (Spergel 1995) Honduras also stands out in the Western Hemisphere in the proportion of its homicides that involve a firearm In 2012 about 84 percent of the homicides involved a firearm (11) The proportion of homicides that involve a firearm appears to be increasing as well In 2008 79 of homicides involved a firearm compared to 81 in 2009 83 in 2010 (United Nations 2013)

71

Over the past decade the nation has responded with ldquoMano Durardquo (ie iron fist or heavy hand)

The new legislation provided the police with more authority to stop search and detain gang

members The new legislation also permitted the courts to sentence gang members to prison

for 12 years for simply being a member of a gang and allowed the courts to sentence

individuals to even longer prison terms for gang related incidents Concomitantly the military

joined the effort to fight gangs by patrolling neighborhoods along side the police While the

public and media strongly supported the shift in national policy toward Mano Dura much of

the evidence suggests that these legislative and policy changes were not effective as the

number of homicides continued to escalate Some suggest that its lack of success was because

gang members who were arrested were released due to of lack of evidence or those who went

to prison if they were not a gang member before entering prison joined a gang Other critics

point out that the heavy handed approach by the government led to loss in the rule of law as

vigilantes engaged in extra-judicial violence against gang members (Seelke 2012) Still others

said that the legislation and policies never really had a chance of working because of the

general lack of effectiveness of the police and courts and the wide spread corruption

throughout the criminal justice system (Zilberg 2011)

As a consequence of the above policymakers and citizens voiced optimism about the possibility

of a truce between gangs after initial results in El Salvador suggested the strategy might be

effective (Villiers-Negroponte 2013) Honduran church leaders and the Organization of

American States (OAS) began to develop a strategy to implement a similar type of truce in

Honduras and the President offered his personal support in their efforts (Arce 2013) In this

case study we examine the processes that lead to the Honduras gang truce and the nationwide

impact of the truce on homicides In the below section we discuss the major stakeholders who

participated in the truce processes leading up to the truce and the establishment of the truce

Key stakeholders

The primary facilitator for the truce process in Honduras was Archbishop Roacutemulo Emiliani who

had earlier served as the Assistant Bishop of the Dioceses of San Pedro Sula and who received

72

support from the Catholic Church to pursue the truce (Bosworth 2013) Prior to the

negotiations he was well known for his work which attempted to establish peace between the

gangs and his advocacy for prison reform and social reintegration programs for gang members

(The Daily Herald 2013) From the onset Monsignor Emiliani proceeded cautiously to ensure

reasonable expectations among the public and policymakers He maintained publically that ldquohe

didnacutet want to be a salesman of false promises about what was going to occur in the future the

things that they do are unpredictable but we expect to have a declaration of reconciliation

principles with societyrdquo (El Mundo 2013) Additionally he wanted to set reasonable

expectations because he knew that it would be a ldquoslow painful and draining processrdquo (El Nuevo

Siglo 2013) and that ldquoWhat is coming is difficult It is not easy It is complicatedrdquo (Castillo

2013)

As in El Salvador the Organization of American States (OAS) played a major role in facilitating

the peace process alongside Monsignor Emiliani Adam Blackwell served as the Secretary of

Multidimensional Security for the OAS and represented Canada on the Honduras Security

Reform Commission (Willcocks 2014) His participation in the mediation process was requested

by Honduran gang members who were in prison They requested that the OAS help broker a

peace agreement with the Honduran government and to help identify resources that would

assist gang members to obtain legitimate jobs (Associated Press 2013) The OAS together with

the Catholic Church served as a ldquobridgerdquo between the executive branch of the government and

the two gangs Additionally two of the mediators (ie Salvadoran Army officer and Police

Chaplain Monsigor Colindres and former Salvadorian congressman Mijango) who helped broker

the truce in El Salvador provided additional support to Monsignor Emiliani and Secretary

Blackwell They traveled to Honduras to present their experiences with the gang truce in El

Salvador and to convey that a gang truce is a promising and legitimate strategy for addressing

gang violence (Associated Press 2013)

Gang leaders of the two primary gangs in Honduras (MS 13 and 18th Street) also participated

extensively in the negotiation process It was stated that they had become weary of the violent

conflict and understood that a truce would be beneficial to the Honduran people (Servellon

73

2013) From the beginning however a number of the critics of the truce argued that Honduran

gangs did not have the capacity to control street level violence They characterized the

Honduran gangs as having less organizational leadership (Bossworth 2013) less control over

turf (LatinNews Daily Report 2013) and being more organizationally fractured (Dudley 2013)

than MS 13 in El Salvador

At the time that discussions about the possibility of a gang truce began President Porfirio Lobo

Sosa was publically supportive of the Catholic Church and OAS negotiating with the gangs

Media reports quoted the President saying I am ldquoprepared to do what ever is necessaryrdquo to

support the mediators (Phillips 2013) ldquoWe have to look for anything thatrsquos an alternative to

violencehellipOn the part of the government we are open to any process that can lower violencerdquo

(Associated Press 2013) and that he had given ldquohis blessing to Emillanirsquos efforts to broker peace

between the gangshelliprdquo (The Daily Herald 2013) However in November 2013 after a general

election the new president Juan Orlando Hernaacutendez through his recently appointed Vice

Minister of Security declared that the government would no longer support the truce process

with the gangs (El Heraldo 2014) Since then the Government of Honduras has not mentioned

the peace process that was initiated in May 2013

Truce making process

It is important to note that prior to the announcement of the gang truce a number of key

stakeholders were somewhat skeptical about its possibility On the one hand some suggested

that a gang truce had been attempted in the past with no success For example one

stakeholder commented to an international media outlet that ldquoEveryone here agrees itrsquos a

positive step forward but people are cautiously optimistic because in 2005 these two gangs

had another peace treaty with each other Now that treaty was very tentative it only lasted

less than two monthsrdquo (Al Jazeera 28 May 2013) On the other hand as noted above other

stakeholders believed that the local gangs did not have enough organizational leadership to

change the behavior of gang members to reduce violence (Bosworth 2013) They argued that

even if gang leaders wanted a gang truce there was no way of enforcing it on the streets

74

Several months prior to the announcement of the gang truce Carlos Mojica Lechuga an 18th

Street Salvadorian gang leader publically stated that representatives of MS13 and 18th Street

in Honduras spoke with several gang truce key stakeholders in El Salvador for the purpose of

replicating the truce in Honduras Tellingly in reflection of the visit Mojica noted that the

advantage of a gang truce is that it formally recognizes Honduran MS13 and 18th Street leaders

as important political persons within the nation He also noted that Honduran gang leaders

have historically been treated poorly and that a gang truce holds the potential for

demonstrating the political power of each of the Honduran gangs (Villiers Negroponte 2013)

The negotiators used different language to describe the early days of the truce Specifically

they mentioned that there had been a consultative process with the gang leaderships so they

were in a process like lighting23 Prior to the truce media sources mentioned that the gang

leaders were offering to stop violence and to not recruit more youth into the gangs24 The gang

leaders also spoke about the ldquopersecutionrdquo they and their family members had suffered during

the previous years highlighting that they had been prohibited social opportunities offered to

others in society (Arce 2013)

Leading up to the negotiations the leadership of both gangs expressed their interest in three

goals 1) lowering violence and crime 2) reconciliation with God society and the government

and 3) helping to improve the social conditions of their communities Although there was not

written documentation on the exact agreement between the parties one MS-13 leader

affirmed that the pact would include all violence (El Comercio 2013) However when talking

about sensitive topics such as extortions which is one of the main sources of income for the

gangs the gang leader said that ldquowould be taken up at a later daterdquo (ABC Internacional 2013)

Leaders of 18th street made similar general statements about ending violence but they were

more specific about their demands One of the 18th Street leaders stated that ldquowhat we want is

23 El Universal 2013 - httpwwweluniversalcominternacional130531obispo-hondureno-descarta-tregua-entre-las-pandillas 24 (Garcia 2013 -httpwwwlaprensahncspmediapoolsitesLaPrensaHondurasSanPedroSulastorycspcid=338546ampsid=276ampfid=98-)

75

to have a dialog with any commission appointed by President Porfirio Lobo and we are sure

that the situation in Honduras will begin to changerdquo (El Nuevo Siglo 2013)

Establishing the gang truce

On May 28th 2013 with public declarations from leaders of both gangs the gang truce was

announced From the beginning of the process the role of the government in the truce was

unclear (La Prensa 2013) Likewise there was little discussion about the exact nature of the

agreement the terms in which gang members would abide and any benefits that would be

made available to those who participated in the truce For example as one stakeholder

indicated the government never decisively considered viable proposals to give the members of

the gangs any opportunities It is important to note that none of the parties signed a formal

commitment and neither MS-13 nor 18th Street signed any type of ceasefire agreement The

gang leaders were in separate locations and were never in direct contact with each other during

the announcement That is both gangs seemingly agreed to the gang truce without ever

talking to each other The ldquopeace processrdquo was publicly announced on May 28 2013 through

ldquojoint but separaterdquo declarations made by the leaders from both gangs imprisoned in the San

Pedro Sula prison (National Penitentiary SPS) (The Daily Herald 2013) The national and

international media widely covered the declarations

The MS-13 leaders said they would not commit any more homicides or any other types of

crimes They ensured that this was an ldquoimmediaterdquo order and would be effective throughout

the country They emphasized ldquoall of the boys know what they have to do starting todayrdquo

(Pachico 2013) The leadership of 18th Street declared that they would stop violence and other

criminal activities but also indicated that the government would have to ldquolisten to themrdquo

Little research has examined whether the gang truce in Honduras ever impacted violence in the

nation Instead anecdotes have been used to portray its effectiveness One facilitator for

example indicated ldquoin Honduras the dialog with the gangs has been positive however the sad

thing in Honduras is that the two main gangs have not accepted a truce between them they

just haven`t accepted it as yetrdquo (La Prensa 2013) despite the fact that ndash in their own words ndash

76

ldquothey do want to hold a dialog with society with the government and with the policerdquo

Conversely gang leaders declared that the truce had been effective For example a leader of

18th Street noted that ldquohellipIt has already done its part telling members in the areas the gang

controls to stop the violence and crimehellip[estimating] crime had already dropped 80 percent in

those areasrdquo (Associated Press June 17 2013) Similarly a member of MS13 estimated that

violence in MS13 controlled areas declined by 45 percent (Associated Press June 17 2013) As

a symbolic gesture of the impact of the truce MS13 leaders also noted that as a gesture of good

will they made and delivered 60 beds for a nursery home in San Pedro Sula (Associated Press

June 17 2013)

Methods

For the present case study we used a pre-test post-test longitudinal quasi experimental design

Data from the 2001 Honduran Census was obtained from the National Institute of Statistics

(INE) These data provided municipal level measures of number of residents population density

per kilometer percent of population who moved in from another municipality ethnicity

percent urban number of residents immigrated to the United States percent female headed

households percent unemployed age composition income percent of households rented and

education level In addition population projections for the years 2005 2010 and 2014 were

also obtained from the National Institute of Statistics (INE) The population levels for the total

population as well as the percentage of residents in a municipality that rural and percentage of

residents who are female were linearly interpolated for the intervening years Examination of

the observed levels indicated that growth was linear overall and so we feel confident that our

linear interpolations are good approximations Second we used municipal level homicide data

by month and year for the period May 2012 through July 2014 These data were provided by

the Honduran National Police through the Honduran US Embassy Both datasets were

merged for the present analysis

Measures

77

The dependent variable for the Honduran case study is the monthly homicide rate which was

calculated by dividing the number of homicides in each municipality by its population and

multiplying this figure by 100000 We examined change by comparing the homicide data 13

months prior to the gang truce to the homicide data 14 months following the truce in each of

the nationrsquos 298 municipalities As presented in Exhibit 21 there were a total of 7910 homicides

over the study period with each municipality averaging 183 homicides (sd=837) A trend

analysis showing the monthly number of homicides on a national level prior to and following

the gang truce is presented in Exhibit 22 It shows that nationally the homicide rate gradually

declined over the study period

Exhibit 21 Summary Statistics

Pre-Truce (n=3809) Post-Truce (n=4101) Total (n=7910)

Mean SD Mean SD Mean SD

Homicide 200 900 166 773 183 837

Population 2874492 8674659 2939557 8870267 2908226 8776124

Homicide rate 521 953 413 814 465 885

Density 9926 10878 9928 10878 9927 10877

Born in the same municipality 8211 1286 8211 1286 8211 1286

Other population group 8545 2390 8545 239 8545 2390

Percent rural 7926 2732 7899 2762 7912 2748

Living in another country 331 650 331 65 331 650

Socioeconomic status 005 097 -002 102 001 100

Exhibit 22 Homicide rate in Honduras by month

78

We used census data to control for several community-level structural factors Principal

components analysis was used to reduce some of the data into a summary measure Exhibit 23

presents the results of the component loadings One component was extracted that we labeled

socio-economic status (SES) This component exhibited high loadings for percent female

headed household percent unemployment and percent completing primary education

Population was used as our exposure variable and population density (per 1 km) residential

stability percentage of the population that is rural percentage of the population that is

indigenous (verify) and the number of residents immigrated to the United States served as our

control variables Population density and outmigration were logged to address skewness in

these two measures

Exhibit 23 Factor Analysis for Socioeconomic Status

Eigenvalue 236093

Variables Factor Loadings

Female-Headed Households 07977

Unemployed 09269

Primary Education 09303

Findings

The results of our t-test analysis are shown in Exhibit 24 It shows that there was a significant

decline in the homicide rate at the municipal level before and after the gang truce The

homicide rate prior to the truce was 697 per 100000 population and following the truce the

homicide rate was 566 homicides per 100000

79

Exhibit 24 T-test on National Data by Month (n= 27 Months)

Homicide Rate Pre Truce Post Truce Change

Mean 6972346 5663592 -1308754

SE 01677361 01151668 02009224

plt 001

Next we examined the effect of the truce through ARIMA regressions The first model in Exhibit

25 does not include an effect of the temporal trend In this model we again find that homicides

declined significantly in the period following the truce We then employed the ARIMA model

again but this time included a variable (month) to control for the temporal trends in the data

Model 2 in Exhibit 25 shows that when we controlled for temporal trends the impact of the

truce we observed was no longer significant Our findings suggest that homicide did not decline

as a consequence of the gang truce but instead the decline in homicides was part of a longer

term historical trend that was independent of the truce

Exhibit 25 Results from the ARIMA Models (n= 27 Months)

Homicide Rate Model 1 Model 2

Effect SE Sig Effect SE Sig

Truce -1309 0350 -0906 0595

m2 -0623 0481

m3 -0153 1447

m4 -0074 0607

m5 -0104 0450

m6 -0050 0466

m7 -0002 0516

m8 0003 0404

m9 0024 1633

m10 -0396 0522

m11 -0754 2424

m12 0837 0366

Time -0030 0047

Intercept 6973 0254 7258 0472

Autocorrelation Coefficient 0010 0408 0111 0249

plt 001

80

Last we used a fixed effects multi-level negative binomial regression to predict the homicide

rate with population as an exposure variable and controlled for the socio-demographic

characteristics of each community The results are presented in Exhibit 26 The only significant

variable in the analysis was the impact of municipal level population stability Specifically we

found that homicide rates increased in areas where residential mobility was high Once again

our analysis showed that time had a negative effect indicating that homicides were decreasing

in all areas over the study period The gang truce itself was unrelated to the decline in

homicides

Exhibit 26 Results from the Multilevel Negative Binomial Models (n= 27 Months)

Homicide Rate Model 1 Model 2

Effect (SE) Sig Effect (SE) Sig

Fixed Effects

Homicide Rate 0004 0004

(0002)

Truce -0119 -0129

(0073) (0072)

Time -0008 -0007

(0005) (0005)

Month 1 0024 0024

(0059) (0059)

Month 2 -0107 -0107

(0062) (0062)

Month 3 -0041 -0041

(0063) (0063)

Month 4 -0033 -0033

(0065) (0065)

Month 5 -0002 -0002

(0067) (0067)

Month 6 (reference)

Month 7 0027 0027

(0051) (0051)

Month 8 0023 0023

(0057) (0057)

Month 9 0036 0036

81

(0056) (0056)

Month 10 -0051 -0051

(0057) (0057)

Month 11 -0031 -0031

(0058) (0058)

Month 12 0187 0187

(0056) (0056)

Born in the same municipality -0013

(0003)

Pecent rural -0003

(0002)

Socioeconomic status 0029

(0044)

Percent dominant population 0001

(0002)

Density (ln) -0010

(0050)

Living in another country (ln) 0001

(0061)

Intercept -10042 -10051

(0061) (0058)

ln(Alpha) -2368 -2367

(0148) (0148)

Random Effects

Var(Truce Effect) 0066 0066

(0024) (0025)

Var(Intercept) 0388 0309

(0049) (0042)

Cov(truce effect intercept) -0024 -0006

(0027) (0025)

plt 001

82

Conclusions

Violent crime in Honduras is widespread With a homicide rate of about 86 per 100000

population Honduras is one of the most violent nations in the world (United Nations 2013)

This compared to an average homicide rate around the world of 62 per 100000 people and

about 27 per 100000 in Central America (United Nations 2013) Given the high rate of violence

in Honduras it is clear that new and innovative ways of reducing violence should be considered

This case study gave an overview of the implementation and impact of the gang truce

experience in Honduras in 2013

The goal of the truce was to significantly reduce the number of homicides In order to assess

the impact of the gang truce in Honduras a pre-test post-test longitudinal quasi-experimental

design was used Census and homicide data were merged at the municipal level to asses

whether the truce had an impact on homicides controlling for population characteristics and

the natural trend in violent crime

Overall the findings suggest that while the homicide rate in Honduras was on a slight

downward trend the gang truce itself was unrelated to any homicide reductions That is the

gang truce had no measurable impact on homicides in Honduras Given the drastic reductions

achieved in El Salvador and the fact that Hondurasrsquo gang truce was a replication of El

Salvadorrsquos the following discussion will examine two important differences between the two

countries gang truces First the implementation of the truce in Honduras did not appear to

obtain trust between all parties involved and did not achieve any notable short-term

deliverables That is the implementation was not robust Second some suggest that the gangs

in Honduras do not have the organizational sophistication to be able to control their members

on the streets rendering them incapable of carrying out any truce agreements

First the implementation of the truce in Honduras did not appear robust The communication

between the parties was weak and none of the parties completed any significant actions as part

of the truce The Catholic Church and OAS served as the bridge between the two main gangs in

Honduras MS13 and 18th Street and the executive branch of the government The ldquopeace

83

processrdquo as it was called in Honduras was initiated in May 2013 After the general election in

November 2013 the newly appointed administration declared that it would no longer support

the truce process with the gangs As a result the truce was short-lived and the governmental

support for the effort shifted with the change in leadership

Though the negotiations were largely based on the good will of a well-respected negotiator

few tangible incentives were offered during the process The negotiations from the beginning

involved discussions about large scale social programming Given the short time frame these

goals in hindsight were unrealistic Naturally there was little trust between the parties at the

beginning of the negotiations and without any quick tangible deliverables from either side the

truce never really materialized

The second challenge to implementing a successful gang truce in Honduras might have been

related to the nature of the gangs themselves It was unclear whether the gangs possessed the

level of cohesion and hierarchical leadership required to make some measures feasible If the

gangs do not have the organizational capacity to control their members on the streets any

agreements that come from the negotiations would be difficult to implement In the earlier

case study of the Salvadorian experience it was suggested that the successes in that country

were related to one of the gangrsquos organizational capacity to impose control of its members The

gangs in Honduras might be different There is at least some evidence that MS 13 in Honduras

might have less organizational leadership (Bossworth 2013) less control over turf (Latin News

Daily Report 2013) and have been more organizationally fractured (Dudley 2013) than their

counterparts in El Salvador The gangrsquos ability to operate as an efficient organization can greatly

impact the outcome of the truce process In general we know that gangs do not have very high

levels of organizational sophistication (Decker Katz and Webb 2008 Decker Bynum Weisel

1998) It might be that gangs in Honduras are more the norm in terms of organizational

capacity

In summary the 2012 truce negotiations in Honduras did not produce any measurable

reductions in the homicide rate The gangs wanted to speak to Honduran society and they even

84

preferred to speak with the governmental authorities but they never interacted with either

The negotiation process seemed to end as quickly as it started The gangs did not deliver with

lower rates of violence and the government did not provide social programs It would be safe to

say that a robust gang truce did not materialize in Honduras The implementation of the truce

seemed to struggle for two primary reasons First the mediators were not able to accomplish

any quick wins to build trust between the parties involved The commitment level on all sides

was not clear throughout the process Second it is not clear whether the gangs in Honduras

have the organizational capacity to control members on the street as would be required to

carry out an effective gang truce In short the 2013 gang truce in Honduras was unsuccessful

85

Conclusions Policy Implications and Recommendations

The purpose of this report was to systematically examine the effectiveness of gang truces Gang

truces have been widely implemented but rarely evaluated Of those gang truces that have

been evaluated little attention has been given to why and how they came into existence In

this report we reviewed prior research on gang truces and presented case studies of gang

truces implemented in El Salvador Jamaica and Honduras for the purpose of understanding

the negotiation processes undertaken with and between gangs and other stakeholders We

were interested in identifying the actors involved in the negotiations the goal(s) of the

negotiations and the strategies employed to carry them out Most importantly however we

wanted to determine whether the gang truce resulted in a reduction in the number of

homicides Each case study offers lessons learned that are unique to their particular

circumstances and when considered together provide direction to policymakers on the

benefits and risks of implementing gang truces

The case studies presented here constitute the most comprehensive evaluations of gang truces

to date Existing documents were used to collect information about the processes associated

with each gang truce Many of these documents included such items as peer reviewed articles

books and reports The majority of these documents were collected over the Internet

requesting documents from those close to the truce and searching library databases Related

the case studies made use of articles obtained from local newspapers The newspaper articles

were not only intended to provide a historical record of the development of each gang truce

but also to provide additional insight into the various external forces that might have impacted

the gang negotiations Because the newspaper serves as a forum for the community to speak

about its concerns newspaper articles also provided a rich source of data on how those in the

community felt about the gang truce Accordingly the newspaper articles offered a different

view of the problem and offered different opinions as to how a gang truce should or should not

be implemented We also conducted a small number of in-depth qualitative interviews with

key informants These data were collected to supplement existing documents and to clarify

issues associated with the negotiation processes This included but was not limited to

86

questions pertaining to identifying the actors involved in the negotiations the goal(s) of the

negotiations and strategies employed to carry out negotiations The interviews were intended

to obtain information from those who possessed first hand knowledge about the gang truce in

each nation

We examined the impact of each truce using official data We first performed a simple t-test

comparing the homicide rates before and after the truce However as discussed above this

technique has limitations The most severe of which is that even if the test was significant it

would be difficult to determine whether the difference was due to the gang truce or a natural

change over time in the outcome We addressed this limitation by using time series models

whereby the homicide rate for the community was modeled as a function of time with truce

period indicators included to measure the effect of the truce net of the temporal trends These

models were estimated with ARIMA techniques Supplemental models were also employed to

examine and control for factors other than the truce that might have impacted homicide over

the study period

Summary of Findings Related to the Implementation of a Gang Truce in the Three Sites

We found that the implementation of gang truces have a number of common characteristics

The first is that in each case a community was experiencing an uncharacteristically high number

of gang related homicides over a fairly lengthy period of time The continued high level of

violence in each case resulted in the community placing strong pressure on the government in

general and the justice system in particular to respond to the problem quickly and effectively

In each case they had first attempted to control gang violence through suppression oriented

strategies and these strategies were found to be ineffective over the intermediate and long

term In turn each communityrsquos inability to exercise traditional informal and formal social

control to decrease levels of violence became self evident to the public and government This

resulted in both the state and community to seek (or participate in) an alternative strategy in

which negotiators would formally andor informally work with gang leaders to establish a truce

that would reduce gang homicide

87

Key stakeholders involved in the negotiation and establishment of each gang truce were fairly

similar In each of the cases examined the gang leaders of the largest and most violently

involved gangs were willing to consider participating in negotiations that could lead to a truce

In each of these cases it was clear that the gangs not only sought to collaborate with the

negotiators for the purpose of reducing violence but perhaps more importantly were seeking a

means in which to gain greater more positive recognition in the community and to reap some

form of benefit to themselves their members and possibly their community In each case

while not always formally involved government officials were at a minimum made aware of

negotiations and in some cases solicited the assistance of third partyrsquos to broker an agreement

between stakeholders In each case it was at least implicitly understood that the government

would ldquolistenrdquo to the gang leaderrsquos expectations of the government and what theymdashthe gang

leaders--had to offer in exchange We found that when the government was no longer willing

to ldquolistenrdquo to or collaborate with negotiators the truce processes ended abruptly Negotiators

were typically comprised of a very small group (ie 2-3) of individuals who were perceived to

be ldquohonest brokersrdquo In El Salvador and Honduras this included a high ranking Catholic Church

official a leader from an international diplomatic organization (ie OAS) and other neutral

parties In Jamaica this included a quasi-governmental organization that had been established

for the purpose of brokering negotiations between gangs to reduce violence and the local

university which had access to staff who were perceived to be neutral but had an interest in

reducing violence due to its proximity to the university

The strategies used to execute each gang truce were similar but yet importantly different They

were similar in that each involved a team of negotiators working to identify common goals to

be achieved and identifying tangibles that could be delivered to the gang leaders gang

members and their community in exchange for the gang achieving their stated goals They

were different however in terms of the structure of the delivery of each parties promise to the

other In Honduras and Jamaica it appears that gang leaders committed to reducing gang

violence in exchange for general promises made by the negotiators for example that

substantial public works programs would be implemented for the goal of reducing

unemployment among gang members and the community In both of these cases it required

88

the government to develop and deploy large scale social programming in a very quick period of

timemdashsomething that neither government had a strong record of demonstrating In El Salvador

negotiators employed a strategy of the gang leaders promising to deliver immediate changes in

gang member behavior for immediate administratively natured changes by the government

For example in exchange for a reduction in gang violence the government agreed to relocate

imprisoned gang leaders to less restrictive prisons and provide them some privileges Following

the successful execution of the first part of the truce which resulted in near term success for

both parties they began to negotiate issues that would take a longer period of time for the

gangs and government to deliver Our findings suggested that some promised deliverables need

to be easily and quickly delivered early in the process so that trust increases between both

parties Stakeholders only have a brief period of time to provide promised benefits before trust

is lost and that tangible benefits need to be delivered in weeks or months not years

Summary of Findings Related to the Impact of Gang Truces in the Three Sites

Our findings suggest that El Salvadorrsquos gang truce had a significant and dramatic impact on the

homicide rate The mean number of monthly homicides declined from about 354 prior to the

truce to about 218 following the truce for a net decrease of about 136 homicides per month

Our forecast showed that between March 2012 and June 2014 the truce had saved about 5501

lives However there was no significant change between the pre-truce and post-truce periods

in the number of thefts extortions robberies rapes and auto theftsrobberies We also found

that the gang truce did not result in a homogenous decline in violence across municipalities

About 61 percent of municipalities experienced a decline in homicides but the decline in

violence varied substantially between municipalities We examined this issue further by parsing

out the relative influence of the number of MS13 and 18th Street gang members on the street

and in prison from each municipality Our analyses indicated that following the truce the

number of MS13 and 18th Street gang members on the street in a municipality was not

significantly related to a decline in homicide but the number of imprisoned MS13 and 18th

Street gang members was associated with a significant change in homicides following the gang

truce In particular the number of imprisoned MS13 gang members was associated with a

89

significant decline in homicides following the gang truce and the number of imprisoned 18th

Street members was associated with a significant increase in homicides following the truce

In Jamaica our initial findings showed that the gang truce might be an effective mechanism for

reducing violence Bivariate analyses showed a significant decline in homicides immediately

after the truce was implemented This explains the work previously published by policymakers

researchers and news reporters Upon further examination of the data however comparing

change in the target and comparison areas and accounting for temporal trends we found that

the decline in homicide was part of a larger nationwide decline in violence and that the gang

truce was not responsible for the decline The only significant effect that we uncovered was the

possibility that homicides were displaced outside the target area for a brief period of time but

then returned to normal

Our findings from Honduras told a similar story as Jamaica Initial analysis showed that the

number of homicides on average declined across municipalities following the gang truce

Specifically the mean number of homicides declined by 13 per 100000 population with 687

homicides per 100000 population on average occurring in each municipality prior to the truce

and 566 homicides per 100000 population on average occurring in each municipality after

the truce However after we examined the effect of the truce through the ARIMA model and

included a variable (month) to control for the temporal trends in the data the impact of the

truce we observed in our bivariate analysis was no longer significant Our findings as in

Jamaica suggested that the decline in homicides was not the consequence of the gang truce

but instead the decline in homicides was part of a long term decline in homicides due to

exogenous factors

The Potential Benefits and Consequences of a Gang Truce

Over the last several years there have been a number of naturally occurring experiments

involving gang truces in a variety of nations in various regions of the world Findings from

evaluations of gang truces are mixed As noted above in El Salvador the gang truce could be

characterized as highly effective at least for the two years following the truce It is worth

mentioning that even after the truce breakup homicides rates while above truce levels

90

continue slightly below pre-truce levels In Jamaica and Honduras the gang truce had no impact

on violence In Los Angeles and Trinidad there was evidence that violence decreased for at

least ninety days but then increased substantially beyond those rates observed prior to the

gang truce (see the introduction section of this report for this discussion) As a consequence it

appears that the potential for long term consequences might out weigh the potential for short

term benefits Only one study site(El Salvador) demonstrated a truce having a substantial and

long term impact on violence Others conversely demonstrated the truce had no impact or

increased violence over the long term In fact a number of scholars have noted that gang truces

are likely to result in a boomerang effect with gang violence increasing over the long run

because of enhanced cohesion within the gang (Klein 1995) Maguire (2013) notes that when

government officials negotiate a truce with gangs they might ldquoinadvertently be acknowledging

gangs as legitimate social entitiesrdquo (p 11) This in itself might increase cohesion among gangs

which has been found to be associated with increased levels of criminality (Decker et al 2008

Klein 1971 Maguire 2013)

Further research is needed to examine how gang truces might impact group cohesion and if it

does whether this in turn results in greater violence Gang truces convey the well-intentioned

image that violence has been addressed and policymakers are doing something about the

problem but researchers need to better understand the probability of a gang truce reducing

violence increasing violence or having no impact This will better position policymakers to

understand the relative risks associated with these types of interventions

Our findings also suggest that while gang truces could be an effective intervention in areas

where gangs are highly structured and organized such as El Salvador they could be counter-

productive in areas where gangs are not as structured and organized Because the vast majority

of street gangs are not well organized (Klein 1995 Spergel 1995) the utility of a gang truce in

reducing violence might be limited Our findings coupled with prior research suggest that gang

interventions need to be tailored to the nature of the gang and its members or it risks

increasing gang violence

Final Thoughts

91

Our analysis suggests that gang truces should only be used as a means of last resort and then

only under certain conditions Given the risks associated with a gang truce communities with

high levels or at least modest levels of formal social control should rely on other more

promising gang control strategies Only when the state has limited or greatly reduced capacity

for social control should a truce be considered Concomitantly a gang truce should only be

considered when a community is experiencing a substantial amounts of gang violence

Communities that are experiencing minimal to modest amount of gang violence may have more

to lose from the establishment of a gang truce than they have to gain Additionally our

findings suggest that a gang truce might only be feasible when gangs are sufficiently organized

to the extent that they have the capacity to regulate memberrsquos behavior In other words gang

leaders must have the ability to reduce their memberrsquos involvement in violence for a gang truce

to work Our findings more concisely suggest that gang truces should only be considered when

there are a great number of gang homicides the state has limited capacity to address the

problem and gang leaders have enough informal social control over their members that they

themselves can substantially control the levels of violence in their community by regulating

their memberrsquos behavior

Recommendations

Gang truces are conjunctural strategies States who suffer from gang-related violence must

establish permanent public policies for crime control and prevention A government that

considers implementing a gang truce should be aware that it cannot become the center

strategy of its public policy for citizen safety

Gang truces should only be used as a means of last resort and then only under certain

conditions Stakeholders must determine whether a process of dialog or negotiation with gangs

is legal ethical and feasible

Stakeholders must anticipate demands that are likely to arise and their response options Some

demands may be easily met such as improved prison conditions Others are much more

difficult and amorphous such as community development through more integrated violence

92

prevention programs (such as those implemented by SolucionES in El Salvador) local economic

development programs or economic reinsertion of ex-gang members

Stakeholders should incorporate immediately achievable and demonstrable deliverables Long-

term goals and promises are unlikely to create the trust needed to sustain a gang truce

Stakeholders must first determine the position in which they are negotiating the incentives that

are possible to deliver and the boundaries and limits they face Gangs are mostly likely to trust

representatives from NGOs community-based organizations and members of the faith based

community as brokers because they are considered more reliably neutral advocates for peace

They need to understand the capacity of the government to deliver promises in a timely

manner

Governments have to make a choice about the visibility and transparency of its participation

This decision needs to be made in the context of the national and local laws the publicrsquos

expectations of transparency and patterns of practices of the past

Governments must be strategic in their support for truce initiatives Some donor funded

programs run by the government prohibit gang member participation and if the government

does not receive approval from the donor it may risk the donor withdrawing its sponsorship of

the program

Governments must ensure an inter-institutional coordination for the management of truces to

avoid the responsibility to be of a single government institution It is necessary to generate or

collect reliable and pertinent data that can be used to analyze and assess the process

It is necessary to implement an effective monitoring system of the truce process similar to that

used in the full report as it can help parties understand what is working and who is delivering

on their promises More specifically the monitoring and truce management system should be

able to identify truce violations and be prepared to respond through the use of legal and

effective practices if stakeholders do not comply

93

Finally it is necessary to develop evaluations of gang truces and monitoring programs and

support violence prevention activities local economic development activities and pilot

programs to support the reinsertion of ex-gang members into society Clearly national

governments municipal governments NGOs and community-based organizations need

increased capacity and resources to discourage the growth of gangs among at-risk youth It is

therefore increasingly important to create economic opportunities for gang members willing to

leave the gangs and find other legal employment Developing and sustaining those

opportunities in nations with high incidences of poverty will require significant international

funding

94

References

Anderson Elijah 1999 Code of the Street New York Norton

mdash 1998 ldquoThe Social Ecology of Youth Violencerdquo In Crime and Justice vol 24 edited by M

Tonry 65-104

mdash 1990 Street Wise Chicago University of Chicago Press

Arce A (2013 May 28) Honduras gangs declare truce ask talks with govt Associated Press

The Big Story Retrieved from

httpbigstoryaporgarticlehonduras-gangs-declare-truce-seek-talk-govt

Archibold Randal 2012 ldquoGangsrsquo Truce Buys El Salvador a Tenuous Peacerdquo The New York Times

Originally published online on August 27 2012

Axelrod R (1986) ldquoAn Evolutionary Approach to Normsrdquo American political science review

80(04) 1095-1111

Ayala Edgard 2012 Gangs Back Plan for Violence-Free Districts in El Salvador

httpwwwipsnewsnet201212gangs-back-plan-for-violence-free-districts-in-el-

salvador Found on November 4 2014

Bakrania Shivit 2013 Policy Responses to Criminal Violence in Latin America and the

Caribbean GSDRC Helpdesk Research Report 934 Birmingham UK GSDRC University

of Birmingham

Bargent James 2013 ldquoRise in Disappearances Feeds Doubts Over El Salvador Trucerdquo August 1

2013 Found at httpwwwinsightcrimeorgnews-briefsrise-in-disappearances-in-el-

salvador-feeds-doubts-over-truce on September 1 2014

Battin Sara R Karl G Hill Robert D Abbot Richard F Catalano and J David Hawkins 1998

ldquoThe Contribution of Gang Membership to Delinquency Beyond Delinquent Friendsrdquo

Criminology 36 93-115

Block Richard 2000 ldquoGang Activity and Overall Levels Of Crimerdquo Journal of Quantitative

Criminology 16 (3) 369-83

iv

Key stakeholders involved in the negotiation and establishment of each gang truce were also

somewhat similar In each case examined the leaders of the largest and most violently

involved gangs were willing to participate In each of these cases it was clear that the gangs

sought to collaborate with the brokers not only for the purpose of reducing violence but

perhaps more importantly as a means to gain greater more positive recognition in the

community and to reap some form of benefit to themselves their members and perhaps their

community In each case while not always formally involved government officials were at a

minimum made aware of negotiations and in some cases solicited the assistance of third

parties to broker an agreement between stakeholders In each case it was at least implicitly

understood that the government would ldquolistenrdquo to the gang leaderrsquos expectations and offers

We found that when the government was no longer willing to ldquolistenrdquo to or collaborate with

brokers the truce processes ended abruptly Brokers were typically comprised of a very small

group (ie 2-3) of individuals who were perceived to be ldquohonest brokersrdquo In El Salvador and

Honduras this included a high ranking Catholic Church official a leader from an international

diplomatic organization (ie OAS) and other neutral parties In Jamaica this included a quasi-

governmental organization that had been established for the purpose of brokering negotiations

between gangs for the purpose of reducing violence as well as the local university which had

access to staff who were perceived to be neutral but had an interest in reducing violence near

the university

The strategies used to execute each gang truce were generally similar but had important

differences Each involved a team of brokers working to identify common goals to be achieved

and tangibles that could be delivered to the gang in exchange for the gang achieving the stated

goals They differed by how each partyrsquos promise was delivered In Honduras and Jamaica gang

leaders traded violence reduction pledges for long-term government and social changes such

as the development of substantial public works programs to reduce unemployment In

Honduras and Jamaica the government was asked to develop and deploy large scale social

programming in a short time framemdashsomething for which neither country had a strong track

record In El Salvador by contrast brokers secured promises for immediate changes in gang

behavior in exchange for feasible immediate deliverables from the government For example in

v

exchange for a reduction in gang violence the government agreed to immediately relocate

imprisoned gang leaders to less restrictive prisons and provide them some privileges Following

the successful execution of the first part of the Salvadoran truce which resulted in near term

success for both parties they began to negotiate broader issues that would take longer for both

sides to deliver Our findings suggest that some promised deliverables need to be easily and

quickly delivered by both parties early in the process to achieve trust and serve as a first test

of gang leadersrsquo ability to deliver Stakeholders have only a brief period of time to provide

promised benefits before trust is lost meaning that tangible benefits need to be delivered in

weeks or months not years

The Impact of a Gang Truce

El Salvador We found that El

Salvadorrsquos gang truce had a definite

impact on the homicide rate The

mean number of monthly homicides

declined from about 354 prior to the

truce to about 218 following the

truce for a net decrease of about

136 homicides per month Our data

show that between March 2012 and

June 2014 the truce had saved

about 5501 lives (see exhibit 1)

From a hypothetical stand it is

possible to make the assumption that a number of these deaths averted could have been

transformed in disappearances and therefore they were not counted within the official

homicide statistics However the results from the analysis point out that the number of

disappearances was not significantly related with the change in the global behavior of

homicides Additionally over the period of analysis (January 2010 to June 2014) there was no

significant change between the pre-truce and post-truce periods in the number of thefts

extortions robberies rapes and auto theftsrobberies

Exhibit 1 Forecast of homicides without gang truce

vi

We also found that the gang truce did not result in a homogenous decline in violence across

municipalities About 61 percent of municipalities experienced a decline in homicides but the

decline in violence varied substantially between municipalities We studied this issue further by

examining the impact of the initiative ldquoFree Violence Municipalitiesrdquo and it was found that the

behavior of violence in those municipalities was not significantly linked to the initiative but

rather to the general dynamic of the truce process nationwide Additionally we parsed out the

relative influence of the number of MS13 and 18th Street gang members on the street and in

prison from each municipality Our analyses indicated that following the truce the number of

MS13 and 18th Street gang members on the street in a municipality was not significantly

related to a decline in homicide but the number of imprisoned MS13 and 18th Street gang

members from the municipality was In particular the number of imprisoned MS13 gang

members from a municipality was associated with a significant decline in homicides in that

municipality following the gang truce and the number of imprisoned 18th Street members

from a municipality was associated with a significant increase in homicides in that

municipality following the truce These findings lend support to the idea that MS13 is more

organized than the typical street gang and that imprisoned MS13 gang members exhibit strong

influence over their fellow gang members on the street Our findings also suggest however

that the gang truce had a boomerang effect in municipalities with high numbers of imprisoned

18th Street members implying that 18th Street might not have as much organizational capacity

to regulate violence on the streets as MS13 The truce provided incarcerated MS13 and 18th

Street gang leaders an opportunity to negotiate with high-ranking officials and influential

diplomats including representatives of the Organization of American States This may have

increased their legitimacy inside and outside of their gangs It appears that MS13 was able to

exert its span of control over the communities in which they had influence and they were able

to deliver on the terms of the gang truce negotiations In the case of 18th Street however

incarcerated gang members may not have had the same organizational capacity for

communicating and carrying out directives In fact a review of the gang truce indicated that

there was a conflict taking place between two factions within 18th Street Consequently the

organizational structure and culture of 18th Street might be more diffuse than that of MS13

vii

and its leadership structure might not be as strong because of the internal fractures within the

gang This might further explain why homicides increased in 18th Street communities The

internal fractures within the 18th Street gang may have resulted in intra-gang violent conflict

that was largely contained within 18th Street controlled territories

Jamaica In Jamaica at first glance our impact findings appeared to show that the gang truce

might be an effective mechanism for reducing violence Bivariate analyses showed a significant

decline in homicides immediately after the truce was implemented This explains the work

previously published by policymakers researchers and news reporters Upon closer

examination of the data however comparing change in the target area to other areas in

Jamaica and accounting for temporal trends we found that the decline in homicide was part of

a larger nationwide decline in violence and that the gang truce was not responsible for the

decline The only significant effect that we uncovered was the possibility that homicides were

displaced outside the target area for a brief period of time but then returned to previous

levels

Honduras Our impact findings from our analysis of data from Honduras tell a very similar story

as Jamaica Initial analysis showed that the number of homicides on average declined across

municipalities following the gang truce Specifically the mean number of homicides declined by

12 per 100000 population from an average of 687 per 100000 population in each

municipality before the truce to an average of 566 thereafter However after we examined the

effect of the truce through time series analysis and included a variable (month) to control for

the temporal trends in the data the impact of the truce we observed in our bivariate analysis

was no longer significant Our findings as in Jamaica suggest that the decline in homicides was

less a consequence of the gang truce than of a broader short to medium term trend And

importantly in both Jamaica and Honduras the respective governments failed to deliver on

gangsrsquo demands for large-scale social and employment programs

Conclusions and Policy Implications

Over the last several years there have been a number of naturally occurring experiments

involving gang truces in a variety of nations and regions Findings from evaluations of gang

viii

truces are mixed In El Salvador the gang truce could be characterized as an effective short

term strategy to reducing homicides It is worth mentioning that despite homicides rates are

above truce levels they continue slightly below pre-truce levels In Jamaica and Honduras the

gang truce had no short or medium-term impact on overall violence In Los Angeles and

Trinidad y Tobago there was evidence that violence decreased for at least ninety days but then

increased substantially beyond those rates observed prior to the gang truce Thus the

effectiveness of the truce in El Salvador appears to be isolated and must be evaluated within

the context of other truces that have failed to reduce homicide violence Policy makers must

evaluate whether the conditions that allowed short term effectiveness of the gang truce in El

Salvador (such as the ability to promise and deliver immediate results) exist in other violent

areas before evaluating whether a truce strategy might be appropriate And they should be

heavily cautioned that the potential for long term negative consequences might outweigh the

potential for short term benefits

Indeed it is important to note that a number of scholars have noted that gang truces are likely

to result in a boomerang effect with gang violence increasing over the long run because of

enhanced cohesion within the gang (Klein 1995) Maguire (2013) notes that when government

officials negotiate a truce with gangs they might ldquoinadvertently be acknowledging gangs as

legitimate social entitiesrdquo (p 11) This in itself might increase cohesion among gangs which has

been found to be associated with increased levels of criminality (Decker et al 2008 Klein 1971

Maguire 2013) Hence it is important to consider the fact that gangs are illegal groups in El

Salvador and it should be cautious when carrying out dialog or negotiation processes with

them Further research is needed to examine how gang truces might impact group cohesion

and if it does whether the cohesion created could be effectively directed toward more

productive non-violent endeavors Gang truces convey the well-intentioned image that violence

has been addressed and policymakers are doing something about the problem but unless the

truce is implemented in a manner and under conditions where immediately achievable results

can be promised delivered and measured there remains a significant chance that the truce will

fail or worse yet backfire Thus it is imperative that any type of concession made by

ix

Governments to gangs within a truce framework should be transparent so that all sectors of

society have certainty that every action is being done within the existing rule of law2

In the case of El Salvador the truce arises from to the absence of effective public policies and

practices for violence control and prevention The truce was planned as a strategy to reduce

gang-related homicides During the process different organizations got involved including

religious (facilitating and protecting human rights) non-government (managing and facilitating

dialog and negotiation processes) international (providing funding for insertion programs) and

government (facilitating and providing certain conditions for dialog and negotiation)

organizations Some of the concessions that the Government provided in order to achieve a

reduction of homicides were within the law but others generated confusion and they seemed

to be close to the legal or socially acceptable limits This fact along with the poor transparency

of authorities in the management of the practice with public media and public weakened the

process and postponed its continuity

The present study suggests that gang truces should only be used as a means of last resort

and then only under certain conditions Given the risks associated with a gang truce

communities with high levels or at least modest levels of formal social control should rely on

other more promising gang control strategies such as pulling levers (ie Boston Ceasefire)

community oriented policing and the Gang Resistance Education and Training (aka GREAT)

program Only when the state has limited or greatly reduced capacity for social control should a

truce be considered Concomitantly a gang truce should be considered as an alternative only

when a community is experiencing a substantial amount of gang violence Communities that

are experiencing minimal to modest amount of gang violence may risk more from the

establishment of a gang truce than they have to gain Additionally our findings suggest that a

gang truce might only be feasible when gangs are sufficiently well organized to be able to

regulate their membersrsquo behavior and cause their members to behave less violently In El

22 Pragmatically this issue is complicated On the one hand transparency is a foundational element within a democracy and is necessary to ensure proper oversight of the government On the other hand it might not be possible to implement a gang truce with too many actors having a voice Policymakers might consider creating a policy that allows such negotiations take place but requires particular actors (such as a judicial body) to be informed of the process to ensure transparency and adherence to the rule of law

x

Salvador there is evidence of the strong organizational structure of gang MS13 Among other

factors the magnitude of its membership the chain of command from its leaders in prison and

the discipline of its leaders in the streets seemed confirmed On the contrary the

organizational structure of gang Barrio 18 ndash divided in two factions fighting over the leadership

ndash showed to be a less stable counterpart within the truce

Finally dialog and negotiations processes with or between gangs must have the capacity to

promise and deliver immediate benefits to the gangs that gang members can see or experience

in order to secure their continued participation in the truce as well as the capacity to monitor

and respond to truce violations Most importantly any effort aimed at reducing violence is

important and should be examined and assessed but it must have a transparent foundation

especially when it affects population rights as a whole

General recommendations

Gang truces are conjunctural strategies States who suffer from gang-related violence must

establish permanent public policies for crime control and prevention A government that

considers implementing a gang truce should be aware that it cannot become the center

strategy of its public policy for citizen safety

Gang truces should only be used as a means of last resort and then only under certain

conditions Stakeholders must determine whether a process of dialog or negotiation with gangs

is legal ethical and feasible

Stakeholders must anticipate demands that are likely to arise and their response options Some

demands may be easily met such as improved prison conditions Others are much more

difficult and amorphous such as community development through more integrated violence

prevention programs (such as those implemented by SolucionES in El Salvador) local economic

development programs or economic reinsertion of ex-gang members

Stakeholders should incorporate immediately achievable and demonstrable deliverables Long-

term goals and promises are unlikely to create the trust needed to sustain a gang truce

Stakeholders must first determine the position in which they are negotiating the incentives that

are possible to deliver and the boundaries and limits they face Gangs are mostly likely to trust

xi

representatives from NGOs community-based organizations and members of the faith based

community as brokers because they are considered more reliably neutral advocates for peace

They need to understand the capacity of the government to deliver promises in a timely

manner

Governments have to make a choice about the visibility and transparency of its participation

This decision needs to be made in the context of the national and local laws the publicrsquos

expectations of transparency and patterns of practices of the past

Governments must be strategic in their support for truce initiatives Some donor funded

programs run by the government prohibit gang member participation and if the government

does not receive approval from the donor it may risk the donor withdrawing its sponsorship of

the program

Governments must ensure an inter-institutional coordination for the management of truces to

avoid the responsibility to be of a single government institution It is necessary to generate or

collect reliable and pertinent data that can be used to analyze and assess the process

It is necessary to implement an effective monitoring system of the truce process similar to that

used in the full report as it can help parties understand what is working and who is delivering

on their promises More specifically the monitoring and truce management system should be

able to identify truce violations and be prepared to respond through the use of legal and

effective practices if stakeholders do not comply

Finally it is necessary to develop evaluations of gang truces and monitoring programs and

support violence prevention activities local economic development activities and pilot

programs to support the reinsertion of ex-gang members into society Clearly national

governments municipal governments NGOs and community-based organizations need

increased capacity and resources to discourage the growth of gangs among at-risk youth It is

therefore increasingly important to create economic opportunities for gang members willing to

leave the gangs and find other legal employment Developing and sustaining those

opportunities in nations with high incidences of poverty will require significant international

funding

xii

References

Cotton Paul 1992 ldquoViolence Decreases with Gang Trucerdquo JAMA 268(4) 443-44

Decker Scott H 2003 Policing Gangs and Youth Violence Belmont CA Wadsworth

Decker Scott H and W Reed 2002 Responding to Gangs Evaluation and Research

Washington DC National Institute of Justice

Decker Scott H Charles M Katz and Vincent J Webb 2008 ldquoUnderstanding the Black Box of

Gang Organization Implications for Involvement in Violent Crime Drug Sales and

Violent Victimizationrdquo Crime and Delinquency 54 153-72

Esbensen Finn-Aage and D W Osgood 1997 ldquoNational Evaluation of GREATrdquo US

Department of Justice Office of Justice Programs National Institute of Justice

Katz Charles M and Webb Vincent J 2006 Policing Gangs in America New York Cambridge

University Press

Klein Malcolm W 1995 The American Street Gang New York Oxford University Press

mdash 1971 Street Gangs and Street Workers Englewood Cliffs NJ Prentice-Hall

Maguire Edward Charles Katz and David Wilson 2013 ldquoThe Effects of a Gang Truce on Gang

Violencerdquo Unpublished paper Washington DC American University

Maguire Edward 2013 Research Theory and Speculation on Gang Truces Woodrow Wilson

International Center for Scholars in Washington DC

McCorkle R and T Miethe 2002 Panic The Social Construction of the Street Gang Problem

Upper Saddle River NJ Prentice-Hall

Ordog Gary J W Shoemaker Jonathan Wasserberger and Michael Bishop 1995 ldquoGunshot

Wounds Seen at a County Hospital Before and After a Riot and Gang Truce Part Twordquo

Journal of Trauma-Injury Infection amp Critical Care 38(3) 417-19

Ordog Gary J Jonathan Wasserberger Julius Ibanez Michael Bishop Eduardo Velayos

Subramaniam Balasubramanium and William Shoemaker 1993 ldquoIncidence of Gunshot

Wounds at a County Hospital Following the Los Angeles Riot and a Gang Trucerdquo Journal

of Trauma 34 779-82

Whitehill J M Webster D W amp Vernick J S 2012 ldquoStreet conflict mediation to prevent

youth violence Conflict characteristics and outcomesrdquo Injury prevention injuryprev-

2012

1

Introduction

Given the increasingly devastating individual family and community effects of gang activity

over the past several decades an increasing body of literature has focused on gangs gang

members and gang activity A core theme running throughout this body of literature is that

gang members are significantly more likely to be the offenders and victims of violent crime than

non-gang members (Battin Hill Abbott Catalano and Hawkins 1998 Esbensen Winfree He

and Taylor 2001 Katz et al 2011 Huff 1998 Curry et al 2002 Miller and Brunson 2000

Miller and Decker 2001 Pyrooz et al 2012 Curry et al 2001) and disproportionately affect

neighborhood levels of crime and violence (Block 2000) These findings have been robust in

that they have been repeatedly found regardless of research methodology (ie use of official

data self-report data observational data) (eg Curry 2000 Decker 1996 Deschenes and

Esbensen 1999 Esbensen et al 2001 Gordon et al 2004 Katz Webb Schaefer 2000 Webb

Katz Decker 2006) or research setting (ie North America Europe Asia South America

Australia) (Klein and Maxson 2006) As a consequence it should not be surprising that

policymakers academics and community leaders have focused much of their attention on

developing responses to address community gang problems

Suppression strategies have been the favored public policy response to gangs since the 1980s

(Venkatesh 1999 Spergel et al 1995) Suppression strategies typically rely on focusing

criminal justice resources on gang members through such practices as targeted and enhanced

police patrols intelligence databases vertical prosecution and enhanced sentences for those

convicted (Katz and Webb 2006) Suppression strategies are based on deterrence theory and

are founded on the principal that swift certain and severe penalties for those involved in gangs

and gang activities will necessarily result in fewer individuals joining gangs and will deter people

from engaging in gang crime (Klein 1995) By the early-to-mid 1990s as gang problems

continued to proliferate and policymakers sought alternative gang control strategies gang

prevention programming flourished (Papachristos 2013) Gang prevention programs were

aimed at the general youth population or focused on at-risk youth or neighborhoods Gang

prevention programming was based on the premise that by reducing risk factors and increasing

2

protective factors prevention specialists could inoculate youth from gang membership

(Esbensen 2000) Prevention programming has most frequently come in the form of school-

based prevention programming such as the Gang Resistance Education and Training (GREAT)

program where students are exposed to a curriculum designed to reduce both participation in

gangs and gang crime as well as the SolucionES3 program which engages youth and their

families through integrated community-based and school based programs

While a burgeoning body of literature has developed describing the assumptions issues and

effectiveness of suppression (eg Decker 2003 Decker and Reed 2002 Katz and Webb 2006

McCorkle and Miethe 2002) and prevention strategies (Esbensen and Osgood 1997) much less

attention has been given to examining gang intervention programming This might largely be

the consequence of the absence of such programming over the past 30 to 40 years By the

1980s policymakers no longer believed that social intervention approaches were an effective

strategy to control gangs and gang problems Although gang intervention strategies took many

forms they were based on the assumption that gang membership is the by-product of a socially

deprived community and that the values and norms of gang youth can be influenced and

directed toward those of mainstream society As such intervention programs often focused on

diverting youth from gangs or sought to minimize the consequences of gangs and gang activity

Gang intervention strategies include crisis intervention dispute resolution street-level

counseling and youth outreach (Spergel 1995) However many policymakers community

organizations and academics have argued that such approaches not only did not reduce gang

activity but may have lead to increased group cohesiveness which in turn may have lead to

increased delinquency (Klein 1971 Spergel 1995) More recent research has yielded similar

results For example a number of studies examining replications of Chicago CeaseFireCure

3 SolucionES is a Global Development Alliance in which $20 million is provided by USAID and SolucionES Alliance

members will contribute an additional $22 million dollars during a 66-month period starting July 2012 SolucionES is being implemented by an Alliance of five leading Salvadoran non-profit organizations who have come together to prevent crime and violence The Alliance members are Fundacioacuten Nacional para el Desarrollo (FUNDE) Fundacioacuten Salvadorentildea para la Salud y el Desarrollo Humano (FUSAL) Fundacioacuten Crisaacutelida (known locally as Glasswing) Fundacioacuten Salvadorentildea para el Desarrollo Econoacutemico y Social (FUSADES) and Fundacioacuten Empresarial para el Desarrollo Educativo (FEPADE) Together these five organizations have notable expertise in education health community development economic development research and youth leadership they are bringing their combined synergy and strengths to prevent crime and violence in El Salvador More information on social violence prevention programming is presented in Annex E including a summary of the violence prevention initiatives being implemented by SolucionES

3

Violence which relies heavily on crisis intervention dispute resolution street-level counseling

and youth outreach have found these strategies to either be ineffective or worse increase

levels of violence (Fox et al 2014)

One type of gang intervention gang truces however has received little attention in the

literature This is somewhat surprising given its use and public claims of effectiveness The

purpose of this report is to systematically examine the effectiveness of gang truces Gang truces

have largely been a ldquoblack boxrdquo that is little is known about the frequency of their use

conditions that give rise to them the role of third parties in brokering gang truces the

transformative effects of truces and the effectiveness of gang truces In this report we review

prior research on gang truces and present evidence on the effectiveness of gang truces that

have been implemented in El Salvador Jamaica and Honduras

Theoretical and Policy Rationales for Gang Truces

Much of the concern about gangs over the past twenty years has been their close association

with violence Qualitative and quantitative research has repeatedly found that gangs and gang

members are involved in high levels of serious violent offending Decker (1996) attributes the

gang-violence relationship in part to the collective and normative structure of gangs which

supports and encourages the use of violence both preemptively and in retaliation He further

explains that gang membership encourages the use of violence in retaliation against threats

and attacks whether actual or perceived which results in patterns of inter-gang conflict

characterized by escalating violence Related Cooney (1998) points out that gangs are not all

that different from ldquowarrior societiesrdquo He argues that while gangs are violent the violence

that they engage in is different than that engaged in by non-gang members in that it commonly

takes on a feud-like dimension A perceived slight violation of turf or other disrespectful

action might invoke a shooting which in turn evokes a retaliatory shooting which in turn

results in another retaliatory shooting creating a self-perpetuating cycle of violence

Early research and theoretical work examining gangs and gang conflict suggested that much

violent gang behavior was the function of status management (Thrasher 1927 Whyte 1943

Short and Strodtbeck 1965) Borrowing from the sociology and social psychology literature on

4

impression management theory gang scholars hypothesized that youth place significant

emphasis on image managementmdashseeking to impress their peers and limiting the potential to

embarrass themselves in front of others (Hughes 2005) These early gang theorists postulated

that violence is an instrument used by gangs and gang members to achieve manage and

protect status They contend that gang members often make decisions to become involved in a

violent conflict based on rational processes that weigh the immediate loss or gain of status

within the gang against the relatively small probability of being formally sanctioned by officials

within the criminal justice system (Hughes 2005) A number of researchers have examined the

relationship between status considerations and gang violence and have found the association

to be particularly robust regardless of gender (Campbell 1991 Giordano 1978) ethnicity

(Anderson 1990 1998 Vigil 1988) and location (Jankowski 1991 Spergel 1995) Hughes

(2005) notes that the centrality of status consciousness by gang youth may account for the

existence of the facilitative gang effect that gang scholars have repeatedly observed (ie Gangs

increase the amount of delinquency individuals are involved in beyond that of a group of

delinquent peers (Gordon et al 2004 Thornberry et al 2003)

Another micro-social factor associated with gang violence is group cohesiveness While our

understanding of the relationship between gang cohesiveness and violence is underdeveloped

some scholars have reported a strong relationship between the two In particular Klein and his

associates (1971 1967) and Lucore (1975) have reported that increases in gang cohesiveness

lengthens periods of gang membership and increases members participation in gang crime and

violence Cooney points out that there are strong relational ties between gang members that

necessarily result in increased cohesiveness among members For example gangs are

comprised of neighborhood youth who share common cultural and economic experiences and

have often known one another for lengthy periods of time He also points out that gangs are

groups that have strong self-proclaimed and formalized identities (eg colors symbols names

monikers) and have at least some organizational structure Their sense of group is maintained

by their common understanding of their members and friends and their attachment to their

territory (or neighborhood) Decker (1996) notes that the relatively high level of group

5

cohesiveness exhibited by gangs facilitates both collective behavior and liability (For exception

see Hughes 2013)

Accordingly both micro-social factors (ie status management gang cohesiveness) serve to

augment levels of gang violence and make it difficult for third parties to intervene Violence

within the context of gangs serves as a form of informal social control Gangs and gang

members cannot seek assistance from legitimate institutions of social control to solve conflicts

because they would risk losing status (Anderson 1999) Likewise the collective nature of the

gang not only increases potential offenders and targets of violence but also facilitates at the

group level the need for retributive justice on the behalf of injured members Moreover gang

members desire to impress others with their commitment to the group and use violence to

demonstrate their commitment to their group and to increase their status within the group All

of this results in an increased cycle of gang conflict and violence4

Over the past several decades residents and policymakers have expected the police to address

violent gang conflicts However there are several limitations to the police response to gangs

First as noted above many of the same factors that are associated with violent gang conflict

also limit the effectiveness of the police to have an impact on violent gang conflict Gang

members are not going to contact the police to resolve a conflict because it could result in loss

of status and expose them to the police discovering illegal activities that they are involved in

(Katz 2001 Katz and Webb 2006) Related citizens in neighborhoods with gang problems are

also reluctant to call the police out of fear of gang reprisals (Katz 1997 Webb and Katz 2003)

or because they have a poor perception of the police themselves (Katz Choate and Webb

2002) Third the police response to gangs in most communities is a reactive response to a

specific incident after it has occurred rather than a proactive response to intervene in ongoing

disputes between gangs (Katz and Webb 2003 2006) Most police agencies simply do not have

4 There is anecdotal evidence for these hypotheses Hughes and Short (2005) through field observations and interviews with members of 20 Chicago gangs examined the micro-social context of violent-related incidents Specifically they were interested in the specific social conditions that precipitated violence Their analysis revealed that gang disputes associated with retaliation escalated into violence significantly more often (ie 55 of the time) than disputes associated with normative order violations or an identity attacks which only resulted in violence roughly 33 percent of time

6

the intelligence networks required to intervene in gang conflicts until after they have risen to

relatively high levels (Katz 2003) Fourth police suppression strategies have been linked to

increasing gang cohesiveness and possible increases in gang crime (Klein 1995)

Some policymakers and community activists have proffered that an alternative to reliance on

formal mechanisms of social control such as the police is the gang truce (Spergel 1995) The

goal of a gang truce is to reduce or even eliminate violent conflict between those gangs that are

warring with one another As such compared with other strategies that often seek to reduce

general levels of crime a gang truce has the very specific goal of reducing violence between

two or more gangs that are in conflict with one another Unfortunately the literature provides

little guidance on the theoretical assumptions of why a gang truce should reduce inter-gang

violence Henderson and Leng (1999) hypothesize that at the root of gang truces is the notion

that they involve the renegotiation of existing norms within and between gangs The authors

claim that as a violent dispute escalates between gangs leaders and members are placed in the

situation of appearing weak to both members of their own gang and to members of the rival

gang if they do not respond with the appropriate amount of force or if they were to suggest a

peaceful resolution to the dispute Accordingly Henderson and Leng argue that as the cycle of

violence escalates between two or more gangs behavioral norms shift toward the increased

valuation of violence to resolve the conflict because it is the only option readily available to

them A gang truce on the other hand which is often mediated by a third-party is believed to

break the cycle of violence by providing the gangs involved in the dispute with a cooling-off

period (Spergel 1995) In the interim period new norms of expected behavior within and

between gangs are established In other words the cooling-off period is believed to recalibrate

norms of behavior that are more consistent with the security interests of the gang and its

members (Henderson and Leng 1999)

For many of the reasons stated above a number of communities have participated in gang

truces Gang truces have been observed in the United States Central America and the

Caribbean and gang leaders in other nations are considering its implantation (Fahah 2012)

Unfortunately little is known about gang truces We do not know when they come into

consideration how they are implemented whether they decrease increase or even have a

7

significant impact on violence In the below section we systematically review the existing body

of literature on gang truces Appendix A describes the methodology we employed to conduct

the systematic review

Prior Research on Gang Truces

Little research to date has examined the effectiveness of gang truces and of the research that

has been conducted has been restricted to the field of public health The most celebrated

truce to date was Philadelphiarsquos House of UMOJA 1974 gang truce In preparation of the truce

gang leaders in prison were called upon for their support along with key community leaders

from churches businesses schools and the police When UMOJA called for the gang summit it

was reported that 500 gang members and 75 of Philadelphia gangs attended By the

conclusion of the gang summit a 60-day truce was announced which reportedly resulted in no

gang members being killed over the 60-day period However no systematic evaluation of the

gang truce was conducted (Woodson 1981)

The first attempt to evaluate a gang truce to our knowledge was conducted by Cotton (1992)

who examined the results of a gang truce in South Central Los Angeles between the Crips and

Bloods Data provided by the police department indicated that over the 6-week period when

the truce took place drive-by shootings decreased by 48 compared to the same 6-week

period the prior year decreasing from 162 to 85 Likewise gang-related homicides dropped by

62 from 26 to 105

Ordog et al (1993 1995) examined the effects of a gang truce in Los Angeles using emergency

room admissions data Specifically the authors examined changes in the daily and monthly

number of gun shoot wound (GSW) emergency room admissions before during and after a

gang truce The catchment area for the emergency room was 100 square miles The authors

noted that while they were able to clearly identify the date the gang truce began because of

the media coverage that it received there was no specified date that the truce ended and as a

5 The authors did not report whether the drop in gang-related homicides was citywide or in the South Central Los Angeles neighborhood where the gang truce took place

8

consequence after 12 weeks the authors regarded it as called off for evaluation purposes

because it was no longer being discussed in the media and gun shot wounds began to increase

Student t-tests were used to examine changes before during and after the gang truce Ordog

et al (1995) reported that there were approximately 7 GSW admissions per day the 12-months

preceding the truce compared to 45 GSW admissions per day during the gang truce and 126

GSW admissions per day in the 11-months following the gang truce The authors concluded

that their analysis ldquoclearly showed that the institution of a gang truce had reduced the number

of GSW victims seen in an hellip inner city Level I trauma centerrdquo (p 419)6

However it is important to point out that while the gang truce in Los Angeles did decrease

homicides by about 35 percent for the first three months it then doubled in months four

through eleven compared with the pre-truce period - a pattern that the authors did not

discuss Similar findings were reported in Trinidad and Tobago (Maguire Katz and Wilson

2013) where it was determined that homicides declined for a brief period of time (again for

about three months) but then increased substantially over the long term (12 months) These

results suggest that gang truces may produce short-term benefits yet result in long-term

adverse consequences

While the research examining gang truces shows their potential promise and their potential for

greater harm we believe that the findings should be viewed with caution for two reasons

First some of the implicit theoretical assumptions that gang truces are built on may not be

accurate Prior research on gangs suggests that they have limited organizational structure

(Decker Katz and Webb 2008 Decker Bynum Weisel 1998) and have few formal

mechanisms to influence member behavior For example most gangs do not have formal

leaders do not require members to pay dues and members do not make contributions to the

gang for the purpose of developing the gangs infrastructure (ie guns housing etc) A number

of academics have pointed out that ldquosophisticated gang organizations is still largely a product of

the self- or organizational-interested musings of gang leaders certain police officials academic

6 Similar results were presented by Ordog et al 1993 using the same data but analyzing it over a shorter period of time and aggregating the data by month instead of by day

9

researchers media reporters based on very limited hard datardquo (Spergel 1995 79-80)

Therefore even if a truce was successfully negotiated between members of gangs in conflict

much of the academic research suggests that gang leaders do not have enough control over

members to enforce a truce Related prior research on gangs and gang members suggests that

they are primarily comprised of young people with few ldquodiplomaticrdquo skills (Henderson and Leng

1999) The very same organizational and normative features of gangs that result in gang

violence (ie often bastions of young minority male street youth) are those same features that

most likely limit the likelihood of a gang having the capacity to abide by a gang truce In sum

implicit in the theoretical assumptions of a gang truce is that gangs have the organizational and

cultural capacity to create and maintain (at least for a short while) a truce which is inconsistent

with much prior research on the organizational characteristics of gangs

The second reason we believe that prior evaluations of gang truces should be viewed with

skepticism is that they have relied on relatively weak evaluation designs For example prior

studies examining gang truces have not incorporated the use of comparison areas or control

groups The causes of reductions in gang violence found in previous evaluations might be

many For example the Los Angeles riots took place just before the gang truce that Ordog et al

evaluated Zinzun (1997) reported that gang culture and violence changed briefly but abruptly

following the riots because gangs and gang members in part redirected their anger and focus

toward the police As such the decline in GSW admissions may have been the consequence of

an overall city-wide decline in gang violence in the wake of the LA riots Related prior research

has relied on fairly broadly defined outcome measures such as GSW admissions and general

levels of drive-by shootings Such outcome measures lack specificity in terms of attributing

gang violence to the specific gangs involved in the gang truce Evaluation designs used in

previous studies have also lacked specificity in terms of the catchment area where violent

activity was labeled as gang related For example Ordog et alrsquos (1993 1995) outcome measure

included all emergency room admissions for GSW However the emergency room received

patients from a 100 mile square area surrounding the hospital an area that was most likely

much larger than the gangsrsquo territories involved in the truce

10

Additionally prior research examining gang truces has not examined the processes involved in

the creation of the gang truce No context has been provided in terms of the factors that lead

to the gang truces whether the gangs were pushed or pulled into truces whether outside

parties helped mediate the truces or whether on-going mediation was required to maintain the

truce In other words we still do not know the processes related to the formation of a

successful gang truce Instead prior research and evaluation has treated the gang truce much

like a black box where it is described in very general terms but its details are not revealed

Likewise prior evaluations have not examined the effects of a gang truce beyond its impact on

gun violence A number of prior researchers have argued that mediating such activities serves

to legitimize gang leaders increases gang identity for members and results in greater group

cohesion (Haskell and Yablonsky 1982 Klein 1995) As such we do not know if there are any

unintended consequences resulting from gang truces due to the methodologies chosen for

prior research and evaluation Finally prior research on gang truces has not relied on more

sophisticated statistical analysis that has the capacity to not only discern overall trends but also

cyclic patterns outliers and turning points

THE PRESENT STUDY

The present study seeks to understand the effectiveness of the gang truce when negotiations

occur with or between gangs and with government institutions or civil society organizations

We have briefly reviewed the theoretical and policy rationales for gang truces and the

literature and prior research on this form of gang intervention Next we present three case

studies one from El Salvador another from Jamaica and another from Honduras Each offers

evidence about their impact on violence and lessons learned that are unique to their particular

circumstances In conclusion the case studies are considered together and we present their

collective policy implications

11

Case Study A Gang Trucemdash The Salvadorian Experience

Over the past two years the Salvadoran gang truce brokered by an array of local policymakers

and international development organizations has attracted national and international

attention Violence in El Salvador has been at an all time high with a rate of 692 homicides per

100000 population in 2011 (UNODC 2014) making El Salvador one of the most violent nations

in the world (Parkinson 2014) Since 1992 government and non-government actors have been

responding to the rising tide of gang violence using traditional suppression strategies such as

increased policing legislative changes and more severe prison sentences These traditional

mechanisms of formal social control however were proving ineffective if not counter-

productive (Perez 2003)

In response to the inadequacy of traditional strategies stakeholders altered their course in an

effort to radically reduce gang violence in the nation Members of the Funes administration led

a group of negotiators comprised of the Catholic Church a former congressman and the

Organization of the American States (OAS) to help frame the conditions for a possible truce

between the MS13 and 18th Street gangs (Umana de Leon and Tager 2014) In March 2012 a

truce was reached The goal of the gang truce was to reduce violence specifically gang-involved

homicides Included in the terms in exchange for the gangs acting to reduce homicides certain

incarcerated gang members were to be transferred to lower security prisons to receive

sentence reductions and special visitation privileges and to be permitted more communication

with those outside of the prison for the purpose of conducting crisis interventions to mitigate

the violence (Salanegra 2012)7 The gang leaders also agreed to no longer recruit children into

their gangs reduce violence against women give up a small number of guns and continue to

participate in negotiations (Seelke 2014 11-12)

7 It is important to note that there were other discussions that took place during the mediations that included as a local publication noted ldquoProhibition Act Gang send the army to barracks end the police operations in controlled areas by gangs repeal of the figure of the witness criteriado (with criteria of opportunity or ldquowitness under a plea agreementrdquo) and a number of improvements in the quality of life of the inmates and its familiesrdquo Source httpwwwsalanegraelfaronetes201209cronicas9612

12

The present case study seeks to understand the role of negotiations with or between gangs and

government institutions or civil society organizations and their impact on violence in El

Salvador The specific objectives of this case study were to (1) Identify and document the

negotiation processes with or between gangs This includes but will not be limited to

identifying the actors involved in the negotiations the goal(s) of the negotiations and

strategies employed to carry out negotiations (2) Identify the impact of negotiations with

andor between gangs on violence and other forms of criminality and (3) Present conclusions

about the strengths and weaknesses of negotiations and their potential for positively or

negatively reducing crime in Salvadoran society

The Salvadorian Gang Truce

While the origins of the Salvadorian gang truce are somewhat unclear some accounts suggest

that in the fall 2011 the Minister of Security (then David Munguia Payes) mentioned the idea of

starting a dialogue between the gangs to Raul Mijango (a former congressman) Shortly

thereafter Mijango spoke to Monseigneur Fabio Colindres the military chaplin about the idea

They started to communicate with gang leaders in January 2012 From the beginning the

Minister of Security stated that he would not personally communicate with gang leaders and

noted that the official position of the government was not to negotiate with offenders But he

did allude to the fact that the gang truce was part of a new strategy to address the nationrsquos

gang problem (Archibold 2012) The Salvadorian gang truce was multi-dimensional involving a

varying number of actors communication styles and tactics In the section below we discuss

the parties involved in the truce the negotiation process and the strategies used by the gangs

and government in furtherance of the truce

Parties involved in the truce There were a series of persons groups and other entities who

directly or indirectly participated in the negotiation process whether they were negotiators or

collaborators From the beginning negotiations between the MS13 and 18th Street gangs were

considered confidential and were held in strict confidence between the gangs and the

negotiators Beginning in early 2012 the first two mediators to serve were a former

congressman (Rauacutel Mijango) and a priest (Fabio Colindres) These individuals were able to gain

13

the trust of the gangs and also had open communications with the government and

community

As the process moved forward however church leadership changed its initial position and

withdrew its support of the truce This left the priest to participate as a private individual no

longer serving as a representative of the church The mediators had some early success

however as the mediators changed the lines of communication became less clear and less

reliable Still they played an important role in the process Finding individuals whom the gangs

would trust and who could speak for government and community institutions was challenging

Further as government ministers and presidents changed it became less clear whether or not

the government was participating in the dialogue between various parties much of the

confusion in negotiations was related to understanding the governmentrsquos official role in the

process

In the beginning the negotiating parties included the two mediators and MS13 and 18th Street

the two predominant gangs in El Salvador MS13 is the largest gang in the country with about

250 cliques throughout the nation For the most part its first- and second-generation leaders

were in prison There are municipalities in which only MS13 operates and therefore its

members never clashed with 18th Street adversaries MS13 members are largely involved in

extortion violence and intimidation of the public The MS13 leadership has been shown to

have greater control of its organization than most its members have been more disciplined

and despite its size the gang seems to have sufficient and effective communication

mechanisms between cliques and its leadership (ldquoranflardquo or ldquoruedardquo)

The 18 Street gang is somewhat smaller than MS13 Long before the truce process began

factions existed within 18th street which had become deeply involved in internal conflicts over

leadership structure Those differences had become more pronounced over time leading to the

establishment of two movements known as the ldquoSurentildeosrdquo (Southerners) and the

ldquoRevolucionariosrdquo (Revolutionaries) that participated in the peace process but under different

14

representation Despite their differences however for the negotiation process the two

fractions of 18th street established similar demands between themselves and with MS138

As noted above the governmentrsquos involvement in the truce process was never clearly

articulated and it varied with ministerial changes The Funes Administration specifically the

Ministry of Justice and Public Security (MJPS) had been engaged in the negotiation process

early on In 2013 however with the change in MJPS leadership this also changed Tensions

grew over the role of the government in the truce negotiations According to the mediators the

new Minister of Security wanted to dismantle the process and for this reason he blocked the

mediatorsrsquo entry to the prisons where they had been meeting with gang leaders The change in

government leadership brought changes in strategy As with all negotiations mutual trust was

the cornerstone When the personnel and their agendas changed so did the trust in the peace

process

Late in 2012 as negotiations began to focus on the relationship between the gangs and the

community other organizations were called upon to help facilitate this interaction A technical

committee was created to oversee progress towards the violence reduction goal this was

announced at the end of August 2012 The committee moved forward with support from the

Organization of American States (OAS) It coordinated actions intended to improve the living

conditions of incarcerated persons to reintegrate into society those who had served their time

in prison and to prevent violence and provide assistance to victims of violence The committee

operated as a mechanism for coordinating the negotiation process with technical support from

the United Nations Development Program (UNDP)‒El Salvador Interviews with experts

indicated that Fundacioacuten Humanitaria with support from several organizations like Interpeace

and others had also supported the operations of the mediators and served as an important

point of reference for the gangs

In each territory where agreements were made government representatives from some

municipalities were invited to participate in the process as key implementation stakeholders

Between December 2012 and January 2013 the municipalities of Ilopango Santa Tecla

8 httpsalanegraelfaronetes201110cronicas5645

15

Sonsonate and Quezaltepeque became among the first to become part of what later became

known as the Sacred Municipalities (aka violence free municipalities) By November 2013

several other municipalities had become part of this group La Libertad Apopa San Vicente

Puerto El Triunfo Nueva Concepcioacuten Ciudad Delgado and Zacatecoluca

The gangs and the government offered several incentives to one another The gangs offered a

number of actions to reduce the incidence of crime (ie cease of hostilities zero homicides

stop extortions robberies and recruitment of children and youths into gangs peace at the

schools) and the government offered to guide public investments in social policies prioritizing

participating municipalities

The truce process was the product of a convergence of vested interests from different sectors

Several earlier attempts at negotiations had failed likely because of a lack of perceived

authenticity on the part of the actors All parties involved in this truce process came to the

table with well-defined group interests and concerns All sides were experiencing fatigue from

the long history of violence Years of fighting deaths and violence had led the mostly

incarcerated first generation those who were still living and who continued to serve as gang

leaders to seek new ways of exercising social power and influence in prison and the

community

The negotiation process In El Salvador the negotiation process between gangs formally began

on March 8 2012 by a decision of the Government to favor the transfer of gang leaders

imprisoned out of the countryrsquos only maximum security prison located in the municipality of

Zacatecoluca That decision was made by the Minister of Justice and Public Security as part of

an agreement between a team of mediators who served as advisors to the Ministry and gang

leaders primarily from MS13 and 18th Street The negotiations were referred to as a ldquotruce

between gangsrdquo and were characterized by the mediators as ldquoa peace processrdquo or a ldquocease of

hostilities between gangsrdquo

Publicly the gang truce was known as an agreement between gangs and not between gangs

and the government because of fear of how those outside the negotiations might interpret the

governmentrsquos actions From the beginning therefore the government never fully

16

acknowledged its participation in the peace process (through representatives) with the gangs

and that reticence was a major impediment throughout the process This was largely because

of the difficulty in communication between the multiple parties which was done through

mediators and necessarily resulted in additional logistical complexity throughout the

negotiations Regardless it is important to recognize that the government played a decisive

and central role in the origin facilitation and promotion of the peace process

In terms of processes interviews with gang leaders indicated that they would communicate

their ldquodemandsrdquo to the mediators who in turn would communicate the gang leaderrsquos demands

to the government The same process was used for the government to communicate with the

gang leaders After an agreement was reached the parties would make proclamations that

emphasized the agreed upon terms of the negotiations These often took the form of public

statements made through local newspapers For example there were more than twenty public

statements made by the gangs where they clearly articulated that their intention was to

reduce the harm that their acts of violence were causing themselves and the communities (For

a detailed discussion of the public statements see Appendix B) In these public statements the

gangs acknowledged the need for their groups to change Both the mediators and the gangs

made clear that the gangs expected the government to respect and respond to their demands

given the good-faith actions that the gangs had already taken

The government on the other hand did not make any formal public statements about their

end of the agreements reached through the negotiations From the outside it appeared that the

government did not develop a formal or consistent policy regarding the truce and instead

assumed that they could achieve a reduction in homicides through the negotiation of increased

prison privileges for gang leaders At the close of the Funes administration the peace process

and negotiations initiated within its framework had reached a plateau There they remained

waiting for the new government to step in and take action as President-elect Salvador Saacutenchez

had announced his support for the truce during the electoral process Sanchez however

withdrew that support when he assumed the presidency in June 2014 Indeed under the

recently elected Sanchez administration the government changed course and became less

17

willing to engage the gangs believing that the government should not negotiate with criminal

organizations

Indeed it should be noted that responses resulting in increased gang influence have been a

concern over the past several decades--from governmental neglect of the problem in the

1990s through the period of the mano dura and super mano dura of the early 1990s to

2000 and to the present gang truce Many have asserted this claim as truth (ie some

responses will increase the influence of gangs) especially in public discourse This study

however does not directly address this point which may or may not be valid9

Strategies employed during negotiations between the gangs and the government Over the

course of the negotiations between the gangs and the gangs and the government at least two

strategies were employed At the beginning of the negotiations strategies were implemented

similar to that of the historic Salvadorian peace accords Each of the parties with the support of

mediators reached agreements and achieved a resolution to their conflict This model

although useful overlooked an important difference between the current violence in El

Salvador and the civil war of the 1980s In a conflict of civilians with the State the legitimacy of

the ldquoadversariesrdquo arises by the need of the state to recover the rule of law It is certainly

possible to reach peace agreements that resolve conflict between gangs but in the current

case with regard to the government there was no legal or policy justification for executing the

truce process On the contrarymdashgovernment negotiations with a criminal group are relatively

rare in modern democratic societies The lack of a legal or policy framework to work from

limited the governmentrsquos ability to be transparent in its response to gangs and may have

undermined its legitimacy with the public

The second strategy employed by the parties was based on reciprocation and cooperation

between actors Early on it was determined by both sides that if one of the parties abandoned

the peace process or did not hold up its end of an agreement the other party would no longer

participate in the negotiation process (Axelrod 1986) In retrospect during the course of the

peace process the gangs implemented agreed-upon terms and positioned themselves as valid

9 We would like to thank one anonymous reviewer for bringing this issue to our attention

18

partners with the government able to negotiate for what is needed by the government in

exchange for what the gang needed10 For example in public statements the gangs insisted

that they had made a goodwill gesture when they declared a unilateral truce and stopped their

involvement in violence The government responded by transferring incarcerated gang leaders

from a maximum-security prison to ordinary criminal prisons11 Next the gangs offered to

disarm the government responded with increased flexibility on control measures at those

prisons This process continued until the demands from both sides grew in a direction that

challenged each sidersquos capacity and willingness to follow through

For instance mediators through the government asked the gangs to end the extortion of

businesses and individuals which are the primary means of subsistence for Salvadorian gangs

The gangs asked that the government reciprocate by eliminating the ldquowitness under a plea

agreementrdquo from the criminal procedural law which is one of the main weapon laws used for

sentencing gang members for complex crimes such as extortion Each of these requests were

more than the parties could carry out Exhibit 1 outlines the different ldquooffersrdquo and ldquodemandsrdquo

made by each party The offers and demands are divided into those that were believed to be

relatively simple and were ldquoofferedrdquo by the party on its own volition and those that were

believed to be more complex and were demanded by the other party

Exhibit 1 Simple demands and concessions made by gangs and the government

Gang concessions Government concessions

Reduce the homicides by more than 50 Transfer the leaders to prisons with lesser security

Hand over 500 firearms to the authorities Allow visits of the children of the gang members

Reduce violent actions at schools Allow night intimate visits

Stop killings at the prisons Allow entry of fast foods

Do not murder custodians police soldiers or their family members

Facilitate the entry into the prison of gang members let out of prison

Give opportunities to some gang members to withdraw

Suspend the searches by the armed forces on persons at the prisons

10 Source httpwwwsalanegraelfaronetes201209cronicas9612 11 This was useful to facilitate the coordination and the communication between the gang members in prisons and in the streets

19

Make it more flexible for new social plans by government to be established in the communities controlled by gangs

Improve the conditions and health assistance at the prisons (eg place tvrsquos to improve mental health of inmates)

Do not burn buses do not kill bus drivers or collecting agents in buses

Allow the Red Cross and journalists to enter into the prisons

The second agenda presented in Exhibit 2 shows the more complex demands that were

not offered by one side or the other but rather were demands placed on the other party These

demands were such that they required a higher level of authority to negotiate in order to

implement the demanded action These demands were considered critical for the peace

process to continue and the delayed responses on both sides stalled the progress of the truce

and led to the parties questioning the legitimacy of the other side all these demands have yet

to be attained The mediators recognized at the end of 2012 that some of the slow progress

was related to the difficulty of making the transition from offers that could be executed

relatively simply to the more complex demands made by each party One of the gang leaders

summarized the issue ldquoWe are not looking for television sets while all of our people continue

living like shit we are not going to try to do everything that is in our power to decrease as much

violence as possible for one television set There are things we are very clear about this [points

to a television set in the room] is a right that the law grantsrdquo

Exhibit 2 Complex demands and concessions made by gangs and the government

Demands by Gangs Demands by Government

Eliminate the figure of witness under a plea agreement from the Criminal Law

Stop homicides and extortions indefinitely

Create international commission to investigate the human rights violation cases of the gang members and their family members on the part of the State (PNC and FA)

Share the information regarding the whereabouts of the brunt of extortion money whether in country or abroad

Institutionalize external and professional surveillance regarding the behavior of the PNC as regards investigations and gang member arrests

Progressive dismantling nation-wide of the clique structures and turfs

Army should stay in its barracks and definitely suspend their participation in public security tasks

Permanently suspend the orders to murder State security and justice agents and their family members (police soldiers judges prosecutors)

20

Clearing of judges prosecutors and police involved in corruption cases against youths in conflict with the law

Permanently suspend the murders extortions and harm caused to public transportation resources and their workers

Maintained a sustained improvement on the conditions of the prisons

Permanently suspend the recruitment of children and youths and hostilities to educational centers

Guarantee working opportunities for the gang members and their family members through specific programs at the municipalities

Share the information about providers of drugs and illegal arms

Television sets were one of the concessions made by the government from the simple list The

gang leader recognized that their intentions in the negotiations extended beyond improving

prison conditions however the government did not have support nor established mechanisms

for carrying out tasks derived from the more demanding list Further as the government

transitioned to new presidential leadership many of the more simple concessions had already

been made and the new government would have to address more complex demands if

negotiations were to continue

The Salvadorian gang truce is remarkable for several reasons First a number of policymakers

and researchers have claimed that the truce saved a large number of lives and was perhaps the

most successful gang truce in the Western Hemisphere Second the Salvadorian gang truce is

somewhat unique in that it might have lasted substantially longer than any other successfully

negotiated gang truce Understanding the temporal impact of the truce is important to

understanding its future potential Third the Salvadorian gang truce is important because a

number of other counties have sought to replicate it For example following the perceived

(and perhaps real) success of the Salvadorian gang truce the nations of Honduras Belize and

Guatemala instituted or tried to institute a similar type of truce Understanding the impact of

the Salvadorian gang truce will further help us understand the potential impact of such a

process on violence in these other nations

Methods

We examined the impact of the gang truce by merging four separate data sets First data from

the 2007 El Salvador Population and Housing Census provided municipal level measures of

21

percent unemployed percent male aged 10-29 percent female headed households ethnic

composition in-migration income percent urban percent households rented and percent

who had a high school education These data were obtained directly from El Salvadors Ministry

of Economy Second municipal level crime data (ie homicide extortion rape theft robbery

and auto theftrobbery) were provided by the National Civil Police (PNC) by month and year for

the period between January 1 2010 and June 30 2014 Disappearance data was also provided

by the police aggregated by year and municipality Third police gang intelligence data was

provided by the National Civil Police (PNC) Specifically we received the number of police

recorded MS 13 and 18th Street gang members by municipality in 2011 the latest year for

which these data were made available Last we acquired 2011 prison gang intelligence data on

the number of imprisoned MS 13 and 18th Street members by municipality from the

Salvadorian National Bureau of Prisons All four data sources were linked together using a

unique municipality identification number and month Collectively they provide an

opportunity to examine the impact of the gang truce in El Salvador and understand several

competing explanations for any changes that might have occurred

The geographic unit of analysis for the present study is the municipality According to the 2007

El Salvador Population and Housing Census there were 14 departments divided into 262

municipalities (the equivalent of a county in the USA) in El Salvador Of the 262 municipalities

ten (38) of the municipalities were eliminated from the analysis because of missing data

Measures

Dependent variables The dependent variable in the present study is the monthly number of

homicides in each El Salvadoran municipality expressed as the number of homicides in

municipality i at time t Homicide data were collected from January 1 2010 through June

30 2014 for a total of 54 months This includes data for a period of 26 months prior to the gang

truce and 28 months following the implementation of the gang truce As shown in Exhibit 3

there were a total of 14148 homicides over the study period with each municipality averaging

371 homicides per month

yit

22

Exhibit 3 Descriptive statistics

n mean sd

Urban Opportunity Factor 252 111 95

male 10-29 years old 252 1960 171

female-headed household 252 3433 499

unemployed 252 1177 686

Racialethnic heterogeneity 252 17 13

of prisoners MS13 252 1579 3960

of prisoners 18th Street 252 1268 4096

of prisoners MS13_spatial weight 252 1579 3960

of prisoners 18th Street _spatial weight 252 1268 4096

MS13 on the street 252 4146 10646

18th Street on the street 252 2513 8622

Violence free municipality 252 04 20

Truce 14148 371 783

Monthly homicide rate per 1000000 14148 51 49

Independent variables Several independent variables were included in our model Our

measure for the intervention is a dummy variable in which the value of 0 is used to represent

the pre-intervention months and 1 represents the post-intervention months We also

included a dummy variable for each community that participated in the violence-free

municipality initiative (sanctuary municipalities) to address issues of external validity (eg

multiple treatment inference) In other words municipalities that participated in the

violence-free initiative might have experienced or participated in something that either

enhanced or detracted from the impact of the larger gang truce The initiative called by some

the second phase of the gang truce took place about 8 to 9 months after the original gang

truce was negotiated in March 2012 In participating sanctuary municipalities gang members

agreed to stop violence and crime in exchange for a reduction in police operations and night

raids12 The municipalities that participated in the initiative included Santa Tecla

Quezaltepeque La Libertad Ilopango Ciudad Delgado Apopa Sonsonate San Vicente

Zacateatecoluca Puerto El Triunfo and Nueva Concepcioacuten

12 See httpwwwcispesorgblogviolence-free-cities-inaugurated-as-second-phase-of-gang-truce

Tt

Si

23

We included several independent variables related to the presence of gangs by gang for each

municipality Specifically we included count variables on the number of MS13 and 18th Street

gang members who were on the streets and in prison by municipality As noted before the

gang truce was made between a relatively small number of imprisoned gang leaders from both

MS13 and both fractions of 18th Street who agreed to stop street-level gang violence on the

condition that they would be transferred to lower security prisons and granted special

privileges

We wanted to understand the impact of the truce in the context of gang presence

Municipalities with high numbers of MS13 and 18th Street members whether they were in

prison or on the streets should have experienced a greater reduction in homicides because of a

greater span of control over these communities Accordingly municipalities with low numbers

of MS13 and 18th Street members should have experienced a lesser reduction in homicides

because of a limited span of control over these communities As Exhibit 3 (above) shows on

average municipalities had about sixteen MS13 and thirteen 18th Street members in prison

and forty-one MS13 and twenty-five 18th Street members on the street We examined whether

gang members in El Salvador were randomly distributed and found that there was spatial

clustering in the number of MS13 and 18th Street members who were in prison In order to

control for the clustering we included the spatial lag (weight) of the MS13 and 18th Street

members in prison

Control variables A series of control variables known through prior research to be related to

violence in communities were added using the census data We began with municipal-level

measures of percent unemployed percent male aged 10-29 percent female-headed

households ethnic composition in-migration income percent urban percent households

rented and percent with a high school education Initial analysis found that five community

variables were highly correlated and loaded on the same factor As seen in Exhibit 4 these

included the percentage of persons who had moved there from another municipality (in-

migration) average income in the municipality percent of the population living in an urban

area percent of houses that are rented and percent of residents with at least a high school

education We labeled this factor Urban Opportunity

24

Exhibit 4 Factor loadings for urban opportunity

Variables Factor

loadings

In Migration Percent of population moved in from another municipality 0638

Income Average monthly income per household (colones) 0886

Urban Percent of population that is urban 0845

Rented Percent of households rented 0761

Education Percent of residents who have at least a high school education 0742

Percent of variance 6071

Eigenvalue 341

Extraction method Principal Axis Factoring

We also calculated a measure of ethnic-heterogeneity from relevant census data Ethnic

heterogeneity which varies from 0 to 1 was calculated by taking one minus the squared

proportions of the population in each ethnic group (White Mestizo other) As with the

percentages of male population that is 10-29 years old female-headed households and

unemployed persons ethnic heterogeneity has consistently been associated with violence in

general (Kubrin and Weitzer 2003) and with gang violence in particular (Katz Maguire and

Roncek 2002 Rosenfeld Bray and Egley 1999)

Analytic Strategy

National-level analysis Our analysis began with a simple t-test of the mean number of

homicides at the national level before and after the truce providing the most basic omnibus

test of an effect We then presented our time series model to estimate the effect net of

seasonality and temporal trends This analysis included a set of simple ARMA (autoregressive

and moving average) models with two lag periods and one period of a moving average Initial

analysis of the number of homicides by month indicated that partial temporal autocorrelations

existed for two lags The first model used only data prior to the truce This model included a

linear time trend variable and dichotomous variables for each month (except January) We used

this model to forecast the expected number of homicides for the truce period This series of

analyses was for illustrative purposes only as the number of data points used in the forecast

25

was too small to provide meaningful confidence intervals The second ARMA model employed

all data from January 2010 through June 2014 and included a dichotomous indicator for the

truce period The effect of this variable was the average change in the number of homicides

net of seasonality (months) and temporal trends

Municipality multilevel models The analysis at the municipal level presented four challenges to

a typical regression model First since our outcome has a highly skewed distribution with

varying levels of exposure (population) we employed a generalized linear model to capture the

correct distribution In this case we employed a negative binomial distribution rather than a

Poisson model due to the over-dispersion in our outcome created by analyzing monthly

municipal datamdashthat is there were several months and municipalities where no or very few

homicides occurred

The second challenge was our need to measure the between-municipality variation of pre-truce

homicide rates and the program effect A fixed-effects model would have been inappropriate

because it would not have allowed us to estimate these variance components Thus we

employed a multilevel random effects generalized linear model

The third challenge was temporal autocorrelation Analysis at the national level indicated

partial autocorrelations in the first and second lags Although a generalized model does not

allow for auto-correlated residuals we addressed this by entering in the model two lags of the

homicide rate

The fourth and final challenge to this estimation was spatial autocorrelation An examination of

the Moranrsquos I and autocorrelation coefficients (exhibit not shown) indicated a low level of

autocorrelation for each month but many months were still statistically significant Thus we

estimated spatial lags of the homicide rate and entered them into the model

Accordingly we analyzed the data using random effects negative binomial models regressing

the rate of homicide on various months on predictors with both temporal and spatial lags For

clarity we present the model using HLM (Bryk et al 1996) notation At the first level (time) we

estimated the log number of homicides using a negative binomial distribution with predictors

26

that included the truce period calendar month linear month trend a

one period lag of the homicide rate a two-period lag of the homicide rate a

spatial lag of the homicide rate and an over-dispersion parameter which has a

Gamma distribution of Thus the final level 1 model is

At the municipality level (level 2) the intercept is a function of the presence of the

Violence Free Municipality program the log number of MS prisoners the log

number of 18th Street prisoners the log of the spatial lag number of MS prisoners

the log of the spatial lag of the number of 18th Street prisoners the log of the

number of MS members the log of the number of 18th Street prisoners the

control variables detailed above and the random effect Thus the level 2 model for the

intercept is

Also at the municipality level (level 2) the effect of truce is a function of SAFE the

log number of MS prisoners the log number of 18th Street prisoners the log of

the spatial lag number of MS prisoners the log of the spatial lag of the number of 18th

Street prisoners the log of the number of MS members the log of the number

of 18th Street prisoners and the random effect Thus the level 2 model for the

truce effect is

TRUCEt mt TIMEt

HRt-1i HRt-2i

HRti vti

Gamma1

aa

aelig

egraveccedilouml

oslashdivide

mti = expb0 i + b1iTRUCEt + lmmt

m=1

11

aring + l12TIMEt

+l13HRt-1i + l14HRt-2i + l15HRti + vti

eacute

euml

ecircecircecirc

ugrave

ucirc

uacuteuacuteuacute

exp vti[ ] ~ Gamma1

aa

aelig

egraveccedilouml

oslashdivide

b0 i

SAFEi PMSi

P18thi

PMSi P18thi

MMSi M18thi

Ci u0 i

b0 i = g 00 +g 01SAFEi +g 02 ln PMSi +1( ) +g 03 ln P18thi +1( ) +g 04 ln PMSi +1( )

+g 05 ln P18thi +1( ) +g 06 ln MMSi +1( ) +g 07 ln M18thi +1( ) + pC

C

aring Ci + u0i

b1i SAFEi

PMSi P18thi

PMSi

P18thi MMSi

M18thi u1i

27

The key parameters of our analysis are in this expression The truce effect is dependent on

the log number of MS prisoners among other characteristics Thus the percent

reduction in homicides can be calculated from this model using the following formula

which we plot below (see Quantitative Findings below)

To answer questions about the variation in the effects of the gang negotiations we predicted

best linear unbiased predictions (BLUP) of the Truce random effect which estimates the

between-community variance of the immediate effect of the truce Since we assumed that the

interventions were not evenly spread throughout communities in this study the mixed model is

appropriate to address this issue We conducted this analysis for homicide using a Stata

generalized linear mixed model using full maximum likelihood and an identity covariance matrix

of random effects

Findings

Exhibits 5 and 6 present our findings related to the number of police recorded homicides in El

Salvador from January 2010 through June 2014 The trends in the data and the results of the t-

test show that in the 26 months prior to the gang truce there were on average about 354

homicides per month compared with about 218 homicides per month in the 28 months

following the gang truce Exhibit 7 shows that the gang truce resulted in 5501 fewer homicides

than otherwise would have occurred

b1i = g 10 +g 11SAFEi +g 12 ln PMSi +1( ) +g 13 ln P18thi +1( ) +g 14 ln PMSi +1( )

+g 15 ln P18thi +1( ) +g 16 ln MMSi +1( ) +g 17 ln M18thi +1( ) + u1i

b1i

ln PMSi +1( )

1- exp g 10 + ln PMSi +1( )g 12( )eacute

eumlugraveucirc( )acute100

u1i

28

Exhibit 6 Number of homicides in El Salvador by month with ARIMA Model Fit

Exhibit 5 Change in homicides from pre-truce to post-truce

Obs Mean SD 95 confidence interval

Pre-truce period 26 35442 4200 33745 37138

Post-truce period 28 21835 6361 19369 24302

Change -13606 10638 16574

=p lt 05

29

Exhibit 7 Forecast of homicides without gang truce

Next as seen in Exhibit 8 we examined the spatial distribution of the change in the homicide

rate after the implementation of the gang truce The analysis showed that of the 252 analyzed

municipalities 243 (93) experienced a decrease in homicides however within these

municipalities there were wide variations in the degree of the decline For example of the 243

municipalities that experienced a decrease the decrease in the homicide rate varied from

about 59 percent of these municipalities experiencing a 1-74 percent decrease to about 9

percent of municipalities experiencing a 75 percent or higher decrease Additionally a modest

number (n=19 7) of municipalities experienced an increase in their homicide rate

30

Exhibit 8 Percent reduction in homicide rate by number of municipalities

Exhibit 9 presents six negative binomial models for the monthly number of homicides The first

model examines the impact of the gang truce and implementation of the Violence Free

Municipality program and our interaction variable that measures the additive effect of both

the gang truce and the Violence Free Municipalities program on the number of homicides We

found that although the gang truce was associated with a significant decline in homicides the

Violence Free Municipality program was related to a significant increase in homicides13

However countrywide the additive effect of implementing the gang truce and the Violence

Free Municipalities program was associated with a significant decline in homicides

13 To be clear our bivariate analyses showed that the violence free municipalities program was related to a significant decline in homicides For example on the one hand those municipalities that did not participate in the violence free municipalities program experienced a decline in their homicide rate from 406 homicides per 100000 population in the pre-truce period to 321 homicides per 100000 in the post-truce period On the other hand those municipalities that participated in the violence free municipalities program experienced a substantially greater decline in their homicide rate from 752 homicides per 100000 population in the pre-truce period to 392 homicides per 100000 in the post-truce period However after other variables are controlled for in our negative binomial models we found that the violence free municipalities program did not have a positive impact on homicides over and above the gang truce itself

31

Exhibit 9 Negative Binomial models for monthly number of homicides

Model 1 Coefficient (se)

Model 2 Coefficient (se)

Model 3 Coefficient (se)

Model 4 Coefficient (se)

Model 5 Coefficient (se)

Model 6 Coefficient (se)

Homicide - 1 month lag 0010 0010 0010 0010 0010 0010

(0002) (0002) (0002) (0002) (0002) (0002)

Homicide - 2 month lag 0012 0013 0013 0013 0013 0013

(0002) (0002) (0002) (0002) (0002) (0002)

Homicide rate spatial lag 0003 0003 0003 0003 0003 0003

(0001) (0001) (0001) (0001) (0001) (0001)

Gang truce implemented -0483 -0541 -0553 0057 -0544 0564

(0066) (0065) (0064) (0278) (0064) (0279)

Violence free municipality 0497 0183 -0003 -0076 -0037 -0094

(0181) (0167) (0152) (0143) (0153) (0145)

Ln( +1) of MS13 prisoners 0066 0176 0062 0171

(0023) (0028) (0043) (0028)

Ln( +1) of 18th St prisoners 0067 0044 0064 -0043

(0021) (0024) (0021) (0024)

MS13 prisoner spatial lag -0004 0054 -0003 0049

(0045) (0055) (0046) (0056)

18th St prisoner spatial lag 0022 0014 0019 0012

(0027) (0032) (0027) (0032)

Gang truce violence free municipality

-0346 -0294 -0309 -0161 -0304 -0059

(0158) (0156) (0158) (0146) (0158) (0145)

Gang truce Ln( +1) of MS13 prisoners

-0185 -0185

32

(0030) (0030)

Gang truce Ln( +1) of 18th St prisoners

0160 0060

(0025) (0025)

Gang truce Ln( +1) of MS13 prisoners spatial lag

-0139 -0138

(0064) (0064)

Gang truce Ln( +1) of 18th St prisoners spatial lag

0023 0023

(0039) (0039)

Ln( +1) of MS13 on street 0023 0018 0021 0016

(0013) (0012) (0013) (0012)

Ln( +1) of 18th St on street 0043 0039 0040 0037

(0016) (0015) (0016) (0015)

Time 0010 0011 0011 0011 0011 0011

(0002) (0002) (0002) (0002) (0002) (0002)

1month 0064 0065 0067 0067 0067 0067

(0055) (0055) (0055) (0055) (0055) (0055)

2month -0020 -0020 -0022 -0023 -0022 -0023

(0056) (0056) (0056) (0056) (0056) (0056)

3month 0088 0088 0188 0088 0088 0088

(0052) (0052) (0052) (0052) (0052) (0052)

4month -0011 -0011 -0011 -0011 -0011 -0011

(0052) (0052) (0052) (0052) (0052) (0052)

5month 0020 0020 0020 0021 0020 0021

(0052) (0052) (0052) (0052) (0052) (0052)

6month 0000 0000 0000 0000 0000 0000

7month 0036 0036 0036 0036 0036 0036

(0052) (0052) (0052) (0052) (0052) (0052)

33

8month 0047 0047 0047 0046 0047 0046

(0051) (0051) (0051) (0051) (0051) (0051)

9month -0055 -0055 -0057 -0058 -0057 -0058

(0053) (0053) (0053) (0053) (0053) (0053)

10month 0040 0040 0041 0041 0041 0041

(0052) (0052) (0052) (0052) (0052) (0052)

11month 0022 0022 0022 0022 0022 0022

(0052) (0052) (0052) (0052) (0052) (0052)

12month -0040 -0040 -0042 -0043 -0042 -0043

(0053) (0053) (0053) (0053) (0053) (0053)

Urban opportunity factor 0209 0042 0033

(0040) (0040) (0039)

male 10-29 0039 0007 0009

(0021) (0019) (0019)

female-headed household -0006 0002 0000

(0007) (0007) (0006)

unemployed 0001 -0002 -0000

(0006) (0005) (0005)

Racialethnic heterogeniety 0058 -0039 -0087

(0263) (0235) (0230)

Intercept -10571

-10544

-11054 -11618 -11015 -11558

(0064) (0060) (0194) (0241) (0212) (0257)

Ln(Alpha) -2287 -2289 -2293 -2294 -2293 -2294

(0113) (0113) (0114) (0114) (0114) (0114)

Truce random effect 0202 0193 0202 0144 0201 0143

(0039) (0038) (0040) (0031) (0040) (0031)

Intercept random effect 0315 0226 0192 0164 0183 0158

(0047) (0035) (0032) (0027) (0031) (0027)

Cov(Truce and Intercept) -0162 -0099 -0122 -0086 -0114 -0090

(0037) (0031) (0031) (0025) (0031) (0026)

34

The second model includes those variables from Model 1 but also includes variables that

controlled for community-level structural factors In this model the gang truce remains

significantly associated with a decline in homicides but the Violence Free Municipality program

and our interaction variable were no longer related to a reduction in homicide Although the

percentage of residents in a municipality who are male aged 10 to 29 female-headed

households percent unemployed and ethnic heterogeneity were unrelated to changes in

homicide the urban opportunity factor was significantly associated with homicide

Models 3 and 4 included our measures of intervention as well as our measures of the number

of MS13 and 18th Street on-the-street and incarcerated gang members at the municipality level

The analysis showed that while the number of MS13 on the street was unrelated to homicides

the number of 18th Street gang members was associated with an increase in homicide These

analyses also indicated that the number of incarcerated MS13 and 18th Street gang members

from a municipality was associated with a significant increase in homicides We further

examined whether this relationship was associated with the gang truce by including two

interaction variables as presented in Model 4 One measured the interaction between the

number of incarcerated MS13 gang members and the gang truce and a second measured the

interaction between the number of incarcerated 18th Street gang members and the gang truce

The results showed that following the gang truce the number of incarcerated MS13 gang

members was related to a significant decrease in homicides and the number of incarcerated

18th Street gang members was related to a significant increase in homicides

Models 5 and 6 in Exhibit 9 present the same two models (3 and 4) as above but include the

community-level structural variables Model 5 once again shows that the number of MS13

members on the street remains unrelated to the change in homicides and the 18th Street

members on the street was associated with a significant increase in homicide However Model

6 shows that following the gang truce the number of incarcerated MS13 gang members

remained significantly associated with a decline in homicides while the number of incarcerated

18th Street gang members remained significantly associated with an increase in homicides

35

Community-level structural factors were found once again to be unrelated to changes in

homicide rates

Exhibit 10 below further illustrates the relationship between reductions in homicides by

municipality and the municipal-level presence of MS13 and 18th Street gang members in

prison The figure shows that in municipalities with no incarcerated MS13 and 18th Street gang

members no change in homicides occurred following the gang truce However when a

municipality had ten MS13 gang members imprisoned on average that municipality

experienced a 55 percent reduction in homicides When a municipality had fifty MS13 gang

members imprisoned those municipalities on average experienced a 71 percent reduction in

homicide following the truce Conversely our analysis showed that the number of 18th Street

gang members in prison (from a municipality) had a significant and positive impact (that is the

number increased) on homicides following the truce For example if a municipality had ten

18th Street gang members imprisoned on average that municipality experienced a 31 percent

increase in homicides14

14 We examined whether there was an interaction effect between gang members on the street following the gangs truce Analysis for both MS13 and 18th street and their interaction with the gang truce showed no effect

36

Exhibit 10 Impact of the gang truce by number of gang members in prison at the municipal level

The Gang Truce and Crime Displacement

We examined two types of displacement crime type displacement and method displacement

Some policymakers have suggested that although homicides may have declined as a

consequence of the gang truce other forms of criminality or categories of reported crime may

have subsequently increased Crime type displacement occurs when offenders who focus on

one type of crime switch to another form of crime in order to avoid detection or to benefit in

some other way (Eck 1993) In El Salvadorrsquos case some critics of the gang truce have suggested

that as homicides decreased other forms of criminality such as extortion might have

increased substantially due to gang membersrsquo increased freedom to conduct activities inside

and outside of prison (Dudley 2013 Parkinson 2014) Method displacement occurs when

offenders change their tactics or methods of conducting crimes as a consequence of an

intervention (Eck 1993) Following the gang truce in El Salvador some analysts argued that gang

37

members might have begun to hide the bodies of homicide victims to avoid detection and to

ldquoprotect the integrity of the countryrsquos gang trucerdquo (Bargent 2013 1)

As noted above we rely on police data that measured monthly numbers of thefts extortions

robberies rapes and auto theftsrobberies by municipality We also received data on the

annual number of disappearances in each municipality by year Exhibit 11 presents the

descriptive statistics for these variables It shows that the rates of theft and robbery did not

change significantly between the pre- and post-true periods and the rate of extortions

significantly declined The t-tests showed that the rates of rape disappearance and

theftrobbery fromof an auto increased significantly

Exhibit 11 Descriptive statistics for measures of displacement

Mean Std Err [95 Conf Interval]

Theft rate

Pre-truce 1238 049 1142 1335

Post-truce 1213 049 1115 1310

Extortions rate

Pre-truce 353 023 308 398

Post-truce 271 021 229 314

Robbery rate

Pre-truce 544 025 495 593

Post-truce 553 025 504 602

Rape rate

Pre-truce 049 003 043 056

Post-truce 064 005 054 073

TheftRobbery Vehicle rate

Pre-truce 132 022 090 175

Post-truce 165 023 120 210

Disappearance rate

Pre-truce 805 068 670 940

Post-truce 1007 064 882 1132

p lt 05

We further examined the data (excepting disappearance data) similarly to the analyses above

in which we used random effects negative binomial models and regressed the number of

38

crimes (ie theft extortion robbery rape and auto theftrobbery) on various months on

predictors with both temporal and spatial lags Our independent and control variables

remained the same as those used in Model 6 (displayed above in Exhibit 9) We found that

over the study period there was no significant change between the pre-truce and post-truce

periods in the number of thefts extortions robberies rapes and auto theftsrobberies (tables

not shown)

As noted above we also received data on the number of disappearances by municipality and

year but because the data were provided by year there were not enough data points to

examine them temporally Therefore we added this covariate (number of disappearances by

municipality and year) to Model 6 in Exhibit 9 The results of the analysis indicated that the

number of disappearances was not significantly related to change in homicides the gang truce

remained associated with a significant reduction in homicides and our interaction variables

(number of incarcerated MS13 amp 18th Street members truce) remained significantly

associated with homicides

Conclusions

The present study sought to examine the impact of the gang truce on violence and other forms

of criminality We examined homicide data by municipality prior to and after the gang truce

Our outcome variables were obtained from the PNC along with several independent and

control variables obtained from the Ministry of Economy and the National Bureau of Prisons

We analyzed this data using a number of time series and random effects negative binomial

models where we regressed the rate of homicide on various months on predictors with both

temporal and spatial lags and controlled for other contextual factors This type of analysis

allowed us to correct for the number of homicides in one month being related to the number of

homicides in the previous month (ie temporal autocorrelation) Additionally it allowed for

the fact that some communities participated in supplemental interventions (ie violence free

municipalities initiative) which might have impacted the outcome in the same way in a given

month These techniques allowed us to isolate the effects of the gang truce as best possible

We also used data from the PNC that measured by municipality the number of thefts

39

extortions robberies rapes auto theftsrobberies and disappearances to examine the

possibility of the trucersquos impact on crime displacement and diffusion of benefits`

Our findings suggest that El Salvadorrsquos gang truce had a significant and dramatic impact on the

homicide rate The mean number of monthly homicides declined from about 354 prior to the

truce to about 218 following the truce for a net decrease of about 136 homicides per month

Our forecast showed that between March 2012 and June 2014 the truce had saved about 5501

lives As will be discussed further below the analysis suggests that key stakeholders have the

capacity to renegotiate existing norms of violence and that at least some gangs have the

capacity to exert substantial informal social control over their members that can result in

reduced violence

The gang truce also lasted substantially longer than previously evaluated truces Ordog et al

(1995) for example reported that the much publicized gang truce in Los Angeles decreased

homicides by about 35 percent for the first three months but then doubled in months four

through eleven compared with the pre-truce period Similar findings were reported in Trinidad

and Tobago (Maguire Katz and Wilson 2013) where it was determined that homicides

declined for a brief period of time (again for about three months) but then increased

substantially over the long term (12 months) These results suggested that gang truces may

produce short-term benefits yet result in long-term adverse consequences

The findings of the present case study suggests that some gang truces might last longer than

previously believed While the number of homicides began to slowly increase about 12 months

following the truce in El Salvador the results clearly showed that almost two years following

the truce homicides still remained below those experienced prior to the truce

We found however that the gang truce did not result in a homogenous decline in violence

across municipalities About 61 percent of municipalities experienced a decline in homicides

but the decline in violence varied substantially between municipalities For example while

about 16 percent of these municipalities experienced a 25 percent reduction in homicides a

number of others (37) experienced a 75 percent or greater reduction Furthermore it is

important to note that about 20 percent of municipalities experienced a modest increase in

40

homicides This suggests that the impact of a gang truce might be variable and could be

dependent on contextual factors We examined the possible influence of these factors by

assessing the impact of social structural factors and the presence of gangs on the municipal-

level impact of the gang truce While we found that social structural factors were unrelated to a

decline in homicides municipal-level gang presence was associated with the decline in violence

as a consequence of the gang truce

We examined this issue further by parsing out the relative influence of the number of MS13

and 18th Street gang members on the street and in prison from each municipality As noted

above we hypothesized that municipalities where gang member presence was high regardless

of their presence on the street or in prison would experience a greater reduction in homicides

because of their increased influence in these areas We found however that the relationship

was not as direct as we would have suspected In particular our analyses indicated that

following the truce the number of MS13 and 18th Street gang members on the street in a

municipality was not significantly related to a decline in homicide but the number of

imprisoned MS13 and 18th Street gang members was associated with a significant change in

homicides following the gang truce

Of special note was our finding of the differential impact of the truce based on gang Although

we found that the number of imprisoned MS13 gang members was associated with a significant

decline in homicides following the gang truce we also found that the number of imprisoned

18th Street members was associated with a significant increase in homicides following the truce

These divergent findings might be associated with each gangrsquos organizational structure and its

capacity to regulate member behavior

Much prior research suggests that imprisoned MS13 gang members have substantial influence

over violence in Salvadoran communities15perhaps even more so than formal mechanisms of

15 Instituto Universitario de Opinioacuten Puacuteblica (Iudop) La situacioacuten de la seguridad y la justicia 2009-2014 entre expectativas de cambio mano dura militar y treguas pandilleras Aguilar Jeannette (Coord) [et al] --1ordf ed --San Salvador El Salv Instituto Universitario de Opinioacuten Puacuteblica (Iudop) 2014- 260 p graacuteficos tablas cuadros y figuras 28 cm (Talleres graacuteficos UCA) httpwwwucaedusviudopwp-contentuploadslibro_la_situaciC3B3n_de_la_seguridadpdf

41

social control such as the police and courts16 Law enforcement officials for years have claimed

that MS13 is one of the most organizationally sophisticated street gangs in the Western

Hemisphere17 The gang has been widely characterized as having a highly vertical organizational

structure and strong control over criminal enterprises in gang-controlled neighborhoods and as

being decidedly capable of enforcing rules through discipline MS13 leadership resides in the

nationrsquos prison system The Ranfla (gang leadership) is comprised of thirteen MS13 gang

members (PNC 2011) who direct coordinate and authorize street crime and other gang activity

from prison Our findings lend support to the idea that MS13 is more organized than the typical

street gang and that imprisoned MS13 gang members exhibit strong influence over their fellow

gang members on the street

Our findings also suggest however that the gang truce had a boomerang effect in

municipalities with high numbers of imprisoned 18th Street members implying that 18th Street

might not have as much organizational capacity to regulate violence on the streets as MS13

The truce provided incarcerated MS13 and 18th Street gang leaders an opportunity to

negotiate with high-ranking officials and influential diplomats This may have increased their

legitimacy inside and outside of their gangs It appears that MS13 was able to exert its span of

control over the communities in which they had influence and they were able to deliver on the

terms of the gang truce negotiations In the case of 18th Street however incarcerated gang

members may not have had the same organizational capacity for communicating and carrying

out directives In fact a review of the gang truce indicated that there was a conflict taking place

between two fractions within 18th Street Consequently the organizational structure and

culture of 18th Street might be more diffuse than that of MS13 and its leadership structure

might not be as strong because of the internal fractures within the gang This might further

explain why homicides increased in 18th Street communities The internal fractures within the

18th Street gang may have resulted in intra-gang violent conflict that was largely contained

16 Instituto Universitario de Opinioacuten Puacuteblica (Iudop) La situacioacuten de la seguridad y la justicia 2009-2014 entre expectativas de cambio mano dura militar y treguas pandilleras Aguilar Jeannette (Coord) [et al] --1ordf ed --San Salvador El Salv Instituto Universitario de Opinioacuten Puacuteblica (Iudop) 2014- 260 p graacuteficos tablas cuadros y figuras 28 cm (Talleres graacuteficos UCA) httpwwwucaedusviudopwp-contentuploadslibro_la_situaciC3B3n_de_la_seguridadpdf 17 httpwwwlaprensagraficacomeua-declara-a-ms-grupo-delictivo-transnacional

42

within 18th Street controlled territories Further analysis is needed to examine this specific

issue

While not the primary focus of this case study we controlled for any impact that the Violence

Free Municipalities program might have had on homicide in 11 municipalities during the truce

As noted above the Violence Free Municipalities program served as the second phase of the

gang truce (Cawley 2013) The program was first proposed by the two mediators and designed

by the Technical Committee on Violence and Crime Reduction which included representatives

from OAS MJPS the mediators and the Humanitarian Foundation (CISPES 2013) The program

involved the mayors of each municipality collaborating with gang leaders to design prevention

and intervention resources for gang members and at-risk youth The Minister of Justice and

Security David Mungia Payes announced that his ministry would facilitate $74 million in

funding from the OAS UN and other donors to implement the programing In exchange gang

leaders agreed to end violence and other criminality in the Violence Free Municipalities (CISPES

2013) The negotiators also agreed to further discuss both gangsrsquo demand to repeal of the

2010 law that increased the capacity of the police and prosecutors to crack down on gangs

(Ayala 2014)

Our findings suggested however that the Violence Free Municipalities program was unrelated

to change in homicides in these communities While much additional research is needed to

understand why the program was not effective it might have been because the gangs had

already agreed to a truce and had already reduced violence to the extent that they could

Conversely the demands made by the gangs as part of the Violence Free Municipalities

program might have been more than could be delivered The time and resources required to

implement the programming and the political capital that was required to repeal legislation

might have been much more than could be delivered Future research is needed to examine the

processes and impact of the Violence Free Municipalities program

Regardless the base rate of violence in El Salvador was reset for a period of time suggesting

that perhaps the gang truce substantially altered existing norms of violence Klein (1995)

argued that cycles of gang violence (that is perceived or real changes in gang activity) are

43

typically the consequence of seasonality epochal variation (peaks and valleys in violence) and

illusion (it appears as if there is a real change in violence but there is not) In El Salvador we

appear to have observed a self-directed cycle of normative change wherein incarcerated MS13

gang leaders directed a reduction in violence by actuating their organizational span of control

through the gang truce Some of the most powerful and influential gang members in El Salvador

used their political social and economic capital to promote the truce and articulate new norms

of violence They were able to do this by leveraging their informal social control over the streets

through actual or perceived threats of violence against those who violated the terms of the

truce

However following a change in government leadership and the government subsequently

distancing itself from the gang truce the conditions and capacity of MS13 leaders to intervene

in local violence might have deteriorated and violence began to increase substantially This is in

part because the third parties in the negotiation were no longer able to communicate with

government officials about furtherance of the truce Specifically the mediators were no longer

able to negotiate on the behalf of the government and were no longer permitted entrance to

the prisons where they could negotiate with gang leaders

44

Case Study B Gang Trucemdash The Jamaican Experience

In the latest report by the UN Office of Drugs and Crime Jamaica ranked as the sixth most

violent nation in the world with a homicide rate of about 521 per 100000 residents (UNDOC

2011) Policymakers and researchers have attributed the nationrsquos high level of violence to such

factors as drug trafficking (Klein Day and Harriott 2004) access to illegal firearms (Lemard and

Hemenway 2006) and historical processes that include a legacy of conflict between the

nationrsquos two primary political parties (Sives 2002 Figueroa and Sives 2003 Moser and Shrader

1999) all of which have facilitated the entrenchment of the more powerful gangs in

communities of the urban poor Likewise high levels of income inequality and chronic youth

unemployment (Francis and Lyare 2006) problematic urbanization (Stone 1975) social

marginalization and an emergent subculture of violence (Harriott 2008) have contributed to the

nationrsquos violence problem Whatever the causes violence has had a considerable impact on

Jamaicarsquos social and economic development it has decreased investor confidence (Schwab and

Porter 2008) tourism (Harriott 2007) and access to public services It has also increased the

costs related to the health care system (Mansinghand and Ramphal 1993) the criminal justice

system (Caribbean Human Development Report 2012) and the education system (Moser and

Holland 1997)

Jamaicarsquos homicide problem is closely associated with its gang problem The Jamaican

Constabulary Force (JCF) has estimated that some 272 gangs are active in the nation most in or

near the capital city of Kingston (Harriott 2014) Gang types and their respective historical

patterns of conflict matter in Jamaica as these variations may determine their predisposition or

amenability to lasting rather than opportunistic truces Jamaican gangs include territorially

organized crime groups conflict gangs defense crews who regard themselves as defenders of

their communities (Levy 2009) and other less cohesive more transient territorial groups Some

of the latter are predatory others bond around the taken identities that generate conflicts of

other similar groups18

18 These are generally rooted in subcultural issues such as the demand to be treated with respect or with the indiscretions and self-centered aggressiveness of individual members that may be associated for example with sexual competition for the favors of women

45

Estimates of the gang-related homicide rate in Jamaica vary perhaps because the crime is not

clearly defined as far as attributing a death to a gang Regardless researchers agree that the

proportion of homicides that are gang related has increased substantially For example Harriott

(2003) reported a fourfold increase in the rate of gang-related homicides between 1983 and

1997 Likewise Hill (2013) using official police data found an eightfold increase from 2001 to

2009 reportedly in 2001 only about 64 percent (n=73) of the nationrsquos 887 homicides were

gang related compared with 523 percent (n=879) of its 1682 homicides in 2009

To address the problem Jamaica has initiated traditional law enforcement strategies such as

establishing a specialized gang unit (Sinclair 2004) initiating curfews (Sinclair and Tuner 2005)

declaring states of emergency (Jamaican Observer 2010) and implementing community-

oriented policing (Kolpack 2006) It also attempted legislative reforms to curb election fraud

and electoral-related violence that involved local gangs (Levy 2009) None of these strategies

stemmed the tide of gang violence In 2002 the Minister of National Security established the

Peace Management Initiative (PMI) (Henry 2011) to augment governmental and non-

governmental organizational capacity to settle gang disputes in the community through

intervention-based programming such as ceasefires and gang truces

The current study examines the peace initiative instituted by the PMI in Greater August Town

Our objective was to understand the negotiation processes undertaken with and between

gangs and other stakeholders Among other things we were interested in identifying the actors

involved in the negotiations the goal(s) of the negotiations and the strategies employed to

carry them out Most importantly however we wanted to determine whether the gang truce

resulted in the desired outcome a reduction in the number of homicides in the Greater August

Town area

The Greater August Town (Jamaica) Peace Initiative

Greater August Town is located on the northeastern outskirts of the city of Kingston The low-

income area has high rates of youth unemployment and a history of gang-related violence

(Charles 2004 Levy 2009) Over the last decade the Greater August Town community has

sought improved living conditions and revitalization (Levy 2009 95) The arearsquos inherent

46

resilience has been augmented by nearby intellectual and cultural engines such as the

University of West Indies the University of Technology and University Hospital (Charles 2004

38)

Greater August Town is comprised of the communities of August Town proper (which is

fractured into several locales with gang-given names such as ldquoVietnamrdquo and ldquoOpen Landrdquo and

city government-approved names such as Hermitage Goldsmith Villa (Angola) and Bedward

Gardens These socially defined community divisions and subdivisions are markers for the

territorial boundaries of street gangs and therefore in some instances are lines of potential

conflict Some of those boundaries demarcate areas of Greater August Town that are

predominantly supportive of one or another political party but the boundaries do not always

hold political significance politics is but one element in the conflict geography of the area Like

many communities of the urban poor the Greater August Town area is easily mobilized

politically ndash a reality that is understood and at times exploited by street gangs who politicize

gang ldquowarsrdquo in their efforts to build alliances and to neutralize the police In fact the basic

principle of community mobilization in Jamaica is political patronage and clientelism Access to

resources (eg jobs housing education) for the poorest residents is often determined by the

local political party Thus according to Charles (2004 36)

supporters attach themselves to the political parties to get first preference in the

distribution of scarce resources and over time because they are unemployable they

become dependent on their political party for their economic survival These supporters

will kill anyone who threatens the support base of their political party because they

perceive it as a threat to their daily survival

As a consequence some political supporters invest heavily in the electoral contests and

provoke conflicts that affirm their loyalty to their party in order to secure material benefits

from it Political competition is one conflict fault line in what otherwise is a politically

heterogeneous community Specifically political support in Greater August Town is divided

between the Peoplesrsquo National Party (PNP) and the Jamaica Labor Party (JLP) The PNP receives

strong support from residents in August Town proper the upper region of Goldsmith Villa and

47

Bedward Gardens while the JLP is supported by those living in Hermitage and the lower region

of Goldsmith Villa (Charles 2004 ) The division between the political parties in the area as in

much of Jamaica was believed to be at the root of much of the communityrsquos violence

particularly between 1979 and 1993 Some gangs aligned with the PNP and others with the JLP

this often resulted in political boundaries overlapping with gang turf (Charles 2004)

Pre-truce Violence

Nationally the history of gang violence in the Greater August Town area first appeared as

political violence closely associated with the electoral cycle As in many other urban

communities the problem intensified mdash deeply affecting community life in the period just prior

to the national elections of 1980 mdash then continued cyclically until 1993 Gangs have since

harnessed this legacy in order to establish and maintain community support based on common

political affiliations They have used that support to nullify the efforts of law enforcement to

suppress their illicit activities (Harriott 2008)

The conflict profile of these gangs and of the communities in which they are nested has

changed over the decades From the beginning of their involvement in political violence and

territorial control a form of gerrymandering existed that manipulated the voting population

forcing certain individuals out of a particular constituency and preventing those who remained

in the community from voting for the opposing party This was linked both to the electoral cycle

and to the types of inter-gang conflicts that typically were associated with street gangs

(Figuerou Harriott and Satchell) More recently however much of the crime and violence

perpetrated by gangs has resulted from internal conflicts (eg status management disputes

over womengirls or money individual membersrsquo activities that could attract police pressure)

Internal conflict at times has led to gang fragmentation and new alliances that pull more parties

into the conflict escalating the homicide rate and increasing the sense of insecurity among the

general population (Levy 2012)

The most significant of these internal conflicts resulted from the killing of former Jungle 12

leader Neil Wright by members of his own gang Jungle 12 was the most influential gang in

Greater August Town Before his murder in order to increase the gangrsquos access to illicit

48

opportunities in Kingston Wright had been attempting to extend Jungle 12rsquos influence with a

system of alliances with other gangs and recruitment in Kingston (Harriott 2014) In short his

ambition was to transform Jungle 12 from a neighborhood street gang operating on the

outskirts of Kingston into to a dominant organized crime network that could reach into the

heart of the city In pursuit of this goal Wright recruited members from outside August Town

elevating them in the gang hierarchy above the locals This led to status-related conflicts and

resistance to Wrightrsquos leadership within the gang His murder precipitated a split of Jungle 12

into three factions two of them fled to other neighborhoods within August Town (Angola and

Vietnam) resulting in the formation of new alliances and a new conflict geography that

replaced the former political geography of conflict Wrightrsquos killing and the subsequent demise

of Jungle 12 as the dominant gang in Greater August Town altered the balance of power and

escalated inter-gang violence (Harriott 2014) The post-2005 phase of conflict was

characterized by power symmetry conflict intensification and the spread of conflict

throughout the entire geographic area of Greater August Town

Although their origins are unclear retaliatory killings and other violent incidents progressively

intensified between 2005 and 2008 The violence was episodic retaliations were most often

motivated by suspicions related to personal and geographic connections between warring

gangs As the violence escalated new alliances were formed to enhance power and dominance

which in turn increased the number of gangs and gang members involved in the violence

(Harriott 2014) This eventually attracted national attention and triggered community

mobilization for a gang truce

The Truce-making Process

The Greater August Town gang truce was preceded by frequent intense violence and public

outrage As noted above the violence had escalated in November 2005 when Jungle 12 leader

Neil Wright was killed The defection of a Jungle 12 member to Goldsmith Villa (Angola) caused

infighting within the gang and conflict between it and Goldsmith Villa Just a few months later

Wrightrsquos brother Steve and two others were injured during a turf battle (Martin-Wilkens 2006)

49

Thereafter violence began to occur at regular intervals until January 2007 when the Peace

Management Initiative (PMI) hosted a peace march in the community Two PNP politicians

urged the community to unite A PMI leader declared that the peace march was being held to

ldquodemonstrate to the public that Jungle 12 members are back together and that they want

peacerdquo (Thompson 2007 1)

Although hopeful some residents remained skeptical about the peace march perceiving the

action to be politically motivated In the absence of trustworthy information inter-group

conflicts tended to be interpreted through a politically partisan lens this created obstacles to

isolating the gangs building a consensus for peace and unifying community mobilization The

politically based narratives weakened the communityrsquos leverage for peace as well as the

exposure of the gangs to police action As one resident said ldquoThe election is coming up and

they want[ed] the people to vote for the PNP is one of the main reasons why they have to walk

todayrdquo (Thompson 2007 1) Those who shared such views stayed away from the peace march

Although that widely held myth was not factual it did serve to demoralize and demobilize one

part of the community A local UWI faculty member was articulate in his assessment of the

politics behind the march

August Town violence is not violence of organized crime which is based on drug

trafficking extortion or some other criminal enterprise [but] rather the violence in

August Town is essentially ldquotribalrdquo mdash the Peoplersquos National Party tribe versus the

Jamaica Labour Party tribe [which has been complicated by a splintering within the

PNP tribe] (2007 1)

Indeed it was not the violence of organized crime but neither was it political violence The

individual quoted above neglected to mention that the conflict was between those pro-PNP

splinter groups who were largely comprised of members of Jungle 12 Moreover their ldquopro-

PNP-nessrdquo was unrelated to the conflict there was no factional infighting within the local PNP

organization at that time

Nevertheless following the peace march the gang violence diminished Then in November

2007 a turf war erupted between two gangs from the Greater August Town neighborhoods of

50

Vietnam and River This time as the police stated the gang violence was less about politics and

more about dominance and turf Police were dispatched to perform directed patrols but

whenever they were not present the shootings continued (Mcleod 2007) In April 2008 the

community witnessed local gangs engaging in a five-hour-long street battle that left two killed

and three others wounded It ended only after the police deployed armored trucks The next

month another round of gang violence resulted in five others being killed including a one-year-

old child This resulted in the three members associated with the gang who committed the

homicides being killed in retaliation (Virtue 2008) The local community mobilized against the

violence increasingly cooperating with the police providing more information about the gangs

Subsequently gang members observed a decline in their influence within the community

During the early period characterized by low-intensity conflict the less influential gangs at

times used manipulation of the police as a tactic for suppressing the more influential gangs

This was largely done through strategic release of information Prior to 2005 when Jungle 12

was dominant its membersrsquo illicit activities were constantly reported to police by members of

other gangs as a means of compelling a compromise or settlement of conflicts In practice this

was done by ldquotrading casesrdquo Once a crime had been investigated by the police and suspects

had been charged an opportunity was created for the gangs and other parties to the conflict to

settle the matter by agreeing to drop their cases (typically by ceasing cooperation with police

investigators) This type of ldquoself-helprdquo served to end some of the retaliations but it rested upon

the manipulation of the police (Harriott 2014)

Later in an attempt to quell escalating inter-gang violence the police established buffer zones

between the warring gangs This action resulted in unintended consequences For example

when the police declared a buffer zone between August Town and Hermitage Hermitage took

advantage of the opportunity to attack Angola Some Angola residents accused the police of

turning a blind eye and creating an opportunity for Hermitage to attack their community

Although little reliable information exists about why the police made the deployment the way

that they did it is more likely that the police inadequately assessed the situation (ie mis-

assessed the pattern of alliances and the likely targets of attack) (Harriott 2014) In the areas

affected by this kind of increasing violence community members became angered and lost

51

confidence in the police The error resulted in some parties to the conflict receiving increased

support from their communities and in greater gang-community cohesion (Harriott 2014)

After a brief period the police identified this problem and began to disengage by no longer

providing a buffer between gang controlled areas which in turn allowed still more conflict to

occur between the gangs

As the violence escalated beyond their control police finally responded by applying their own

forms of pressure For example units under the direction of the JCF High Command would

make periodic raids in the community during which they would at times seize weapons and

make mass arrests (Sinclair 2005) However there were also moments when the local police

were very responsive improved their relations with the community and consequently gained

greater access to relevant information Two such moments occurred just prior to and again

immediately after the truce moments during which there was greater freedom of movement

and open collaboration between the community and the police (Harriott 2014)19

The Establishment of the Greater August Town Gang Truce

The Greater August Town gang truce was led by the Jamaican Peace Management Initiative

(PMI) The PMI is a government-funded initiative created for the purpose of working with gang

members to reduce violence Due to community mistrust of the police in 2002 the organization

was established as an alternative organizational mechanism for responding to gang violence

The PMI sought to bridge government and civil society efforts to mediate disputes between

gangs as well as to provide outreach to gang members (Bakrania 2013) While efforts to

institute a gang truce in Greater August Town were led by the PMI a number of other

stakeholders helped to facilitate the truce these included faculty at the University of the West

Indies (UWI) and representatives from the police the faith-based community and the August

Town Sports and Community Development Foundation (Jackson 2008 Levy 2009 also see

Appendix C) The gangs involved in the truce included those from August Town Hermitage

19 The quality of police-community relations largely depended on the style of the local station commander however regardless of the external environment

52

Goldsmith Villa Bedward Gardens and African Gardens Because of its formality as well as its

perceived effectiveness the truce signed on June 24 2008 was regarded by many as the first

of its kind in Jamaica (Levy 2009)

Truce negotiations began early in June 2008 and lasted for about three weeks The gangs

sought to leverage their violence-making capabilities and demanded payment for peace They

asked the third-party negotiators for money ldquoworkrdquo and start-up funds for proposed micro-

businesses (Wilson 2014) Those demands were rejected by the negotiators on the grounds that

the third-party institutions would not buy a peace that was intended to save the lives of those

who were making the demands Moreover if peace was to be purchased then gang conflict

could be used continuously to extract money and other benefits from negotiators The third-

party actors made some demands of their own In some quarters of the community and society

the surrender of guns was viewed as a litmus test of the sincerity of the gangs Consistent with

this the negotiators suggested that the parties to the conflict symbolically hand over one gun

each that suggestion was immediately rejected by the gang leaders These kinds of demands

from the various parties ceased after a time as they all agreed that the truce was to stand on

its own merits (Harriott 2014)

As the truce began to be committed to paper a number of stakeholders expressed concern that

their greatest risk in participating could be the potential for Jungle 12 factions to use the peace

agreement as they had in the past as a tactic to persuade their enemies to let their guard

down Others however recognized that Jungle 12 had now been weakened and that a formal

public peace agreement would be beneficial to the gang and therefore this time would be

different (Harriott 2014)

The gangs held fast to their claim that their weapons were needed for their own protection

because the police were ineffective in responding to violence in their communities (Jamaican

Gleaner 2014) It became a precondition of the truce that the gangs would not be required to

turn in their guns and other weapons (Jackson 2008) The truce agreement did specify

however that ldquoall persons are allowed to move freely across all boundaries regardless of

reputation or affiliation No gun salute or any other shooting is to take place in the community

53

for a period of at least five yearsrdquo (2008 also see Appendix C) The truce agreement and its

conditions were prescribed in a document that was finally signed by all of the major

stakeholders including the gangs (see Appendix C)

Throughout the negotiations each of the gang leaders had attempted numerous times to use

the truce as an opportunity to bargain for money jobs and business support grants Such

demands consistently were rejected by the third-party actors Nonetheless both prior to and

after the truce some efforts were made to create better opportunities for young people

residing in the community UWI for example provided a homework supervision program to

encourage students to further their education and it developed a community-building initiative

to help improve schools and enhance sports programming (Levy 2009) Such programs were

conducted as part of UWIrsquos Township Project in August Town which invested significant

resources in developing the residentsrsquo job-related capabilities and collective self-efficacy

The Greater August Town gang truce was noteworthy for two reasons First the gang truce

received substantial press attention The media were invited to witness the ldquosigningrdquo of the

truce by the gang leaders in the presence of a JCF Deputy Commissioner of Police a PMI board

member and the UWI Principal and two professors of its faculty Second the truce was widely

credited with decreasing violence in Greater August Town and it served as an exemplar to

other communities seeking to replicate its success (Virtue 2008) A number of reports

manuscripts and newspaper articles proclaimed the truce to be a success Bakrania (2013 10)

for example reported that ldquoPMI has been credited with stopping gang wars in August Town

rdquo Levy (2009 94) remarked that the ldquomost interesting outcome of PMI efforts to date was the

Peace Agreement reached in August Town in late 2008rdquo Likewise a government report noted

that ldquothe peace treaty was a pivotal achievement in August Town that has significant potential

for wider application Crime levels dropped markedly in August Town after the signing of the

peace agreement in June 2008rdquo (McLean and Blake-Lobban 2009 78) To this day August Town

celebrates the signing of the truce with an annual celebration with food and music

(Cunningham 2011)

54

Methods

Our evaluation relied on a pre-testpost-test quasi-experimental group design Our

methodology examines the Greater August Town community which is comprised of three

contiguous towns where the gang truce took place (the target area) and the balance of Jamaica

which is comprised of 178 communities (comparison areas) As seen in Exhibit 12 the average

number of residents living in each of the three communities in the target area was not

significantly different than that for the rest of Jamaica about 7776 residents lived in each of

the Greater August Town communities compared with 6468 in the other communities

Likewise communities of Greater August Town were about as densely populated as other

communities (2960 per square kilometer versus 2647 per square kilometer) and the age

range of residents was similar as well However Greater August Town (a) had a significantly

higher proportion of its residents living in poverty (196 vs 158) (b) consumed fewer

resources than other communities and (c) reported significantly more homicides than other

communities prior to the truce (see Exhibit 12)

55

Exhibit 12 Descriptive characteristics of Greater August Town and balance of Jamaica (2007-2011)

Comparison

Area Greater August

Town All areas

Population (mean) 6468 777633 648994

(sd) 720482 353731 715621

Population density (mean) 264719 296033 265238

(sd) 271023 285501 270465

Percent in poverty 1577 1957 1583

(sd) 1036 106 1029

Consumption 15737890 1106939 1566048

(sd) 10713020 205336 1064021

residents under 15 yrs old 2369 2494 2371

(sd) 487 115 484

residents 15-65 yrs old 6840 6901 6841

(sd) 423 29 419

Murder per month (mean) 674 857 677

sd 1928 1409 1920

Total murders 10068 180 10248

n 178 3 181

plt=05

Measures

Two distinct data sets were merged to measure the impact of the Greater August Town truce

First data from the 2011 decennial census provided community-level measures of the social

and economic characteristics of the 181 communities in Jamaica Described in detail below the

community-level data used in the study included population population density gender age

poverty and consumption20 These data were obtained directly from the Statistical Institute of

Jamaica

Second police homicide data from the years 2007 through 2011 were used to construct the

studyrsquos community-level measure of homicide The homicide data were aggregated by month

20 Consumption is an alternative measure of poverty in Jamaica which measures the consumption of food and non-food items

56

and appended to the community-level data The final (merged) data set included 10248

homicides over the 60-month study period These data were obtained from the Jamaica

Constabulary Force (JCF)

The dependent variable examined in the study was constructed from official police homicide

data Once again the homicide data represented the number of officially recognized homicides

in Greater August Town and each of the remaining communities in Jamaica We examined

change by comparing the homicide data 18 months prior to the truce with the homicide data 42

months following the truce More specifically we examined whether there was a change in the

number of homicides in the 30 days following the truce (month 1) as well as whether the truce

had an impact every three months thereafter (ie months 2-5 6-8 9-11 12-14 15-42) and

whether any changes in homicide coincided with changes in homicide in the balance of

observation areas The frequency distribution of our dependent variable is presented in Exhibit

13 It shows that prior to the truce the target area on average experienced significantly more

homicides (1495) than did the comparison areas (920)

57

Exhibit 13 Distribution of homicides in the target and comparison areas

Comparison

Area Target

Area Total

Pre-truce period Mean 920 1495 932

SD 2469 1966 2461

N 241400 5100 246500

Month 1 of truce Mean 741 286 733

SD 1785 496 1772

N 17800 300 18100

Months 2 thru 5 of truce Mean 647 905 652

SD 1821 1249 1812

N 71200 1200 72400

Months 6 thru 8 of truce Mean 577 1236 588

SD 1690 1074 1683

N 53400 900 54300

Months 9 through 11 of truce Mean 718 333 711

SD 2034 733 2019

N 53400 900 54300

Months 12 through 14 of the truce Mean 687 095 678

SD 1519 286 1509

N 53400 900 54300

Months 15 thru 42 of the truce Mean 564 589 564

SD 1683 1042 1674

N 516200 8700 524900

Total Mean 674 857 677

SD 1928 1409 1920

N 1006800 18000 1024800

An illustration of the trends in homicide prior to and following the gang truce are shown in

Exhibit 14 It shows that 30 days following the truce homicides fell in the target and

comparison areas then increased and decreased several times with a general downward slope

in violence over time

58

Exhibit 14 Monthly number of homicides pre-post truce in the target and comparison areas

We also used a number of measures to control for community-level structure from the 2011

decennial census These community-level data included the communityrsquos population

population density (per square kilometer) and community level of consumption Additionally

the census data included measures of the percentage of the population that was female under

15 years old 15 and 65 years old and 65 years old and older as well as a measure of the

percentage of the population living in poverty Principal components analysis was used to

reduce some of these data into a summary measure

Exhibit 15 shows the results of the component loadings One component was extracted that we

designated as socio-economic status (SES) which exhibited high loadings for percent living in

poverty percent under 15 years old percent 15 to 65 years old and consumption Excluded

from the principal components analysis were population and population density Population

was used as our exposure variable and population density was logged to address skewness in

these data

Exhibit 15 Factor loadings from principal components factor analysis

Loading Poverty 78 Consumption -76 under 15 years old 92 between 15 and 65 years old -80

59

Analytic Strategy

In order to test whether the truce had an impact on homicides in the target area andor

whether displacement had occurred in the balance of the study area several analytic

techniques were employed Most of the methods employed the use of the homicide rate as the

dependent variable We explored the data in this way to provide the maximum statistical

power to detect an effect As a check on these methods we also employed a generalized model

to compensate for the non-normality of our outcomes

First focusing only in the target area we performed a simple t-test comparing the homicide

rates before and after the truce (the unit of analysis was a month) however this technique had

limitations the most severe of which was that even if the test were significant it would be

difficult to determine whether the difference was due to the gang truce or a natural change

over time in the outcome Second to address this limitation time series models were

employed whereby the homicide rate for the target area was modeled as a function of time

with truce period indicators included to measure the effect of the truce net of the temporal

trends These models were estimated with ARIMA techniques with a one-month lag auto-

correlated error Third we examined the homicide rate for each town using a panel time series

model In this model the temporal trend for each town was examined with indicators for target

areas and truce periods included The main effects for the truce periods measured the effect of

the truce in the target areas and the moderators of the truce period in the comparison areas

measured displacement effects Finally because the dependent variable coded is not normally

distributed across months we used a negative binomial time series model to estimate the

number of homicides with the population covariate serving as an exposure variable

Findings

The first set of results examines only the target area The first test was a simple t-test

comparing the mean homicide rates before and after the truce periods The result was a mean

difference in the homicide rate of -890 per 100000 with a significant t-statistic of 370 While

60

this result is statistically significant we caution that it may or may not reflect an impact of the

truce To further examine the truce effect in the target area we performed ARIMA regressions

The first model did not include an effect of the temporal trend In this model we again find that

by month 15 the murder rate decreased by about -89 per 100000 (Exhibit 16)

Exhibit 16 Results of basic ARIMA model

Next we employed the ARIMA model again but included a variable (date) to control for the

temporal trends in the data Exhibit 17 shows that when we controlled for temporal trends the

impact of the truce we observed was no longer significant This result indicates that it was not

the truce per se that caused the decline in homicides but instead the decline in homicides was

part of a larger (local and nationwide) decline in homicides

Note The test of the variance against zero is one sided and the two-sided sigma 7877339 5907266 1334 0000 6719536 9035142 L1 1720758 1242218 139 0166 -0713944 415546 ar ARMA _cons 1538436 1904285 808 0000 1165203 1911669 t_15 -8888017 2969383 -299 0003 -147079 -3068134 t_12 -1507749 1820406 -083 0408 -507568 2060181 t_9 -132475 9189969 -144 0149 -3125951 4764507 t_6 -2861008 6335492 -045 0652 -1527835 9556328 t_2 -5166363 4498142 -115 0251 -1398256 3649834 t_1 -1111771 4463623 -025 0803 -9860311 7636769murders_rate murders_rate Coef Std Err z Pgt|z| [95 Conf Interval] OPG

Log likelihood = -2089969 Prob gt chi2 = 00397 Wald chi2(7) = 1472Sample 564 - 623 Number of obs = 60

ARIMA regression

61

Exhibit 17 Results of ARIMA model with control of temporal trends

We next estimated the possible displacement effects of the truce Exhibit 18 presents the

results of these models Examination of the main effects of the truce period does not indicate

any effects and looking at the truceComparison interaction effects we also do not find any

displacement effects Note that these models also controlled for the socio-demographic

characteristics of each community

Note The test of the variance against zero is one sided and the two-sided sigma 7775398 6622814 1174 0000 647735 9073446 L1 1270618 1253989 101 0311 -1187156 3728392 ar ARMA _cons 1480274 1197324 124 0216 -8664378 3826986 date -2318963 2091772 -111 0268 -6418761 1780835 t_15 -3519687 7694691 -005 0964 -1543329 1472935 t_12 -970747 2228007 -044 0663 -5337561 3396067 t_9 -8844212 9402792 -094 0347 -2727335 9584921 t_6 2995697 6576644 005 0964 -1259042 1318956 t_2 -2353533 4923798 -048 0633 -12004 7296933 t_1 -9194625 5462116 -017 0866 -1162501 9786088murders_rate murders_rate Coef Std Err z Pgt|z| [95 Conf Interval] OPG

Log likelihood = -2082031 Prob gt chi2 = 00710 Wald chi2(8) = 1444Sample 564 - 623 Number of obs = 60

ARIMA regression

62

Exhibit 18 Results of Panel (Town) time series model with control of temporal trends

Last we used a random effects negative binomial regression that predicted the homicide rate

with population as an exposure variable and controlled for the socio-demographic

characteristics of each community The results are presented in Exhibit 19 The analysis showed

that time had a negative effect indicating that homicides were decreasing in all areas over the

study period The main effects of the truce (truce = 1 2 ) represented the effects of the

truce in the targeted area and did not show a significant effect for any period following the

gang truce However we did find that the homicide rate significantly increased in the

_cons 6310151 3518738 179 0073 -5864483 1320675lnpopulationdensity 121961 1709831 713 0000 8844891 1554731 ses 1245334 2428629 513 0000 7693312 1721336 date -1276623 0345659 -369 0000 -1954101 -0599144 15Balance 5392222 4128863 131 0192 -2700202 1348464 12Balance 1318516 7903159 167 0095 -2304746 2867507 9Balance 1042658 7925876 132 0188 -5107854 2596101 6Balance -1703928 7919006 -022 0830 -1722489 1381704 2Balance 3498997 7175155 049 0626 -1056405 1756204 1Balance 1056874 1104696 096 0339 -1108291 3222039 truceaugust Balance -4757021 3276618 -145 0147 -1117907 1665032 august 15 -4269255 4284086 -100 0319 -1266591 41274 12 -1262496 7871907 -160 0109 -2805361 2803697 9 -1014276 7881965 -129 0198 -2559113 5305603 6 6563019 786731 008 0934 -1476334 1607595 2 -5034368 7123478 -071 0480 -1899613 8927392 1 -1215158 1095699 -111 0267 -3362688 9323715 truce murders_rate Coef Std Err z Pgt|z| [95 Conf Interval]

Prob gt chi2 = 00000 Wald chi2(16) = 12359 max = 60 avg = 5661878Estimated coefficients = 17 Obs per group min = 48Estimated autocorrelations = 1 Number of groups = 181Estimated covariances = 1 Number of obs = 10248

Correlation common AR(1) coefficient for all panels (02045)Panels homoskedasticCoefficients generalized least squares

Cross-sectional time-series FGLS regression

63

comparison areas in months 12 through 14 following the truce In particular we found a 29

percent increase in the homicide rate in the comparison communities for that period (exp (-

1797 + 2048) = 1285 p-value = 004) Since this effect is only significant at the 005 level

however and given the number of analyses used to examine the data it is possible that we

found this effect by chance alone

64

Exhibit 19 Random Effects Negative Binomial

65

Given these caveats we visualized this model with the following set of marginal predictions as

observed in Exhibit 20 We saw that the targeted area (as illustrated in red) experienced an

immediate decrease in homicide which coincided with an increase in homicides in the balance

of the study area However the target area quickly returned to ldquonormalrdquo and homicides in the

comparison area decreased again During months 9 through 11 following the truce there was a

reduction in homicides in the target area with an associated increase in the comparison area It

is important to point out that the confidence intervals are large and we cannot yield concrete

conclusions from these results However it appears that the truce might have had a temporary

short-lived displacement effect decreasing homicides in the target area but increasing

homicides in the comparison area

Exhibit 20 Predicted change in homicides in the target and comparison areas

Conclusions

From 2000 through 2009 Jamaica experienced a substantial number of homicides many of

which were attributed to gangs in one form or another Traditional law enforcement responses

were repeatedly implemented but until 2010 those had little effect Some policymakers in

Jamaica as well as in other nations throughout the Caribbean and Central America have

recently been experimenting with novel approaches to reducing gang-related violence notably

the implementation of gang truces In Jamaica at least eight gang truces reportedly have been

66

negotiated since 2001 (Levy 2009) The Greater August Town gang truce was thought to have

been one of the more successful and it has served as a model for other communities to use

(2009) Our purpose here has been to identify the actors involved in the negotiations of that

truce the negotiation goals and the implementation methods used and then to examine

empirically the impact of that truce on homicide rates in the targeted community

The 2008 gang truce in August Town was a response to violence that arose when the leader of

one gang was killed creating a power vacuum that other gangs saw as an opportunity to

increase their influence in the community Concomitantly the community as well as the gangs

feared that an absence of formal social control would result in further violence The police

reacted unevenly At some times they engaged in appropriate but heightened levels of

preventive patrol while at other times they purposely provided little or no protection on

occasion they used aggressive tactics that further isolated them from the community The end

result was that there was neither stability nor predictability in the police response and

therefore little trust in the police to address the problem

As the violence further escalated the community mobilized The Jamaican Peace Management

Initiative faculty members from the University of the West Indies (UWI) the Jamaican

Constabulary Force (JFC) and other community-oriented groups joined forces seeking to

reduce the increasing number of homicides by brokering a truce between the gangs Over the

three-week negotiation period the negotiators and the gangs sought terms from one another

The gangs wanted payment ldquoworkrdquo and funds for micro-business development to end the

violence The third-party stakeholders wanted the gangs to disarm actually or symbolically

Neither the gangs nor the stakeholders had substantial leverage nor did they have much to

offer one another in terms of incentives In the end however a truce was agreed upon and all

of the gangsrsquo leaders and several key community stakeholders signed it at a public ceremony

with the media in attendance

At first glance our impact findings appeared to show that the gang truce was an effective

mechanism for reducing violence Bivariate analyses showed a significant decline in homicides

after the truce was implemented This explained the work previously published by

67

policymakers researchers and news reporters Upon further examination of the data however

comparing change in the target and comparison areas and accounting for temporal trends we

found that the decline in homicide was part of a larger nationwide decline in violence and that

the gang truce was not responsible for the decline The only significant effect that we

uncovered was that possibly the homicides were displaced outside the target area for a brief

period of time but then returned to normal

Any one of a number of explanations might be offered for the strategyrsquos lack of effectiveness It

might be that the Jamaican gang leaders at least those in Greater August Town did not have

the organizational capacity to change gang member behavior Much prior research suggests

that in general gangs have limited organizational structure and little formal leadership This

might suggest that gangs do not possess the necessary capacity to regulate their membersrsquo

violence That said gangs in Jamaica including in Greater August Town have been found to be

fairly organizationally sophisticated and to possess strong leadership

In fact in a small number of Jamaican communities gangs have been found to be highly

organized with individual gang leaders being referred to as dons and community leaders The

gang leader in such a community is often found to have substantial control over members and

residents as these communities often turn to the don rather than the police for justice The

don will hold court and punish those who commit crime Punishment can include beatings and

torture as well as execution (Morgensen 2004) Although this level of organizational structure

and sophistication is found only in a small number of Jamaican communities generally the

gangs in Jamaica are believed to have some organizational capacity or at least enough to

reduce violence in communities

Our findings however indicated that prior to 2005 and the death of Neil Wright perhaps only

Jungle 12 could approximate that capacity to discipline members and enforce a truce After the

gangrsquos fragmentation in 2005 Jungle 12 lost much of its organizational capability and

enforcement of the truce was therefore difficult The truce negotiators sought to address the

enforcement issue by proposing a peace council that would involve all parties The proposal

was approved by all key stakeholders still some gang leaders demanded cash payments as a

68

condition for attending council meetings Peace was consistently seen by then as a bargaining

tool rather than as an honest attempt to establish and maintain peace In the end members of

only two gangs were attending the meetings21 and the council soon dissolved

In an effort to replicate the council function UWI sponsored one of the most respected

negotiators a community activist to become a one-person monitoring and intervention

specialist or a ldquoviolence interrupterrdquo His job was to ensure that truce violations did not lead to

a return of the gang wars mdash and there were many violations of the truce For example there

were instances of gang members crossing boundaries and entering the turf of another gang

armed although not initiating conflict behavior that was interpreted by the opposing gangs as

preparation for the next round of ldquowarrdquo or as laying a foundation for a surprise attack that

would exploit the truce for this purpose In the absence of the council these matters were

reported to the violence interrupter who tried to resolve the problems in consultation with the

various gang leaders Often the gang leaders were unresponsive or incapable and therefore the

threatening practices and violence continued Ultimately there were no rules or bodies or

persons who could regulate the violence and there were never any reference points for

compliance The formal truce agreement was an attempt to negotiate and impose such rules

via a collective pressure that would include third parties but it was unsuccessful in doing so

The potential for re-engineering norms related to conflict thus was not realized

Another explanation for the failure of the gang truce might be that it was more a vehicle for

rhetoric rather than for reality The gang leaders insisted that they would sign the truce

agreement only if it were ratified in public with the presence of the media (Jackson 2008 Levy

2009) The leaders might have viewed the process in and of itself as a means of increasing their

reputation and influence within the community and in policymaking circles (and to reduce

mutual distrust) In signing the truce gang leaders publicly pledged to reduce their involvement

in violence thereby calming local residentsrsquo fears They also made public efforts to increase

resources for their communities perhaps in an attempt to portray themselves as ldquoprovidersrdquo to

the community In fact the truce did provide gang leaders with an opportunity to be seen in

21 Interestingly the Jungle 12 factions did not attend any of the peace council meetings

69

public collaborating with important community stakeholders The imagery of the public signing

was of the government (via the PMI) and others approaching the gang to ask them to use their

means of informal social control in the community to reduce violence mdash to accomplish

something that the government could not do on its own As a consequence the process may

have been perceived by gang leaders as a victory because it enhanced the gangsrsquo reputation

with both the government and community

Alternatively from the start the gangs might not have been fully invested in the gang truce

One of the major criticisms of the Greater August Town gang truce was that gangs were not

required to give up their firearms although some believed that this was an unrealistic request

their demand and the demand of many that all guns be turned in immediately was

quite unrealistic given the decades of ingrained gun culture and the continued inability

of the security forces to guarantee protection for any corner against armed rivals It was

obvious to most observers that that kind of situation could not be ended overnight and

that this was a reasonable first step in the process (Levy 2009 63)

The gangs feared that if they were to disarm themselves they would be vulnerable to other

gangs and unable to protect themselves a concern that appears not to have been addressed by

mediators Indeed at times some elements within the community felt somewhat dependent on

the gangs to maintain security If the gangs would have been disarmed and there were no

near-term alternative prospects for any form of social control both the gang and the

community might have faced additional violence as has been observed in the past In the end

the gang truce only called for a reduction in gang violence and did not provide any solutions to

address the larger problems between the gangs nor did it provide the gangs with any tangible

benefits for abiding by the truce

70

Case Study C Gang Trucemdash The Honduran Experience

Introduction

Violence in Honduras is at epidemic levels increasing almost 44 percent over the past five

years In 2012 there were 7172 homicides in Honduras or about 86 homicides per 100000

population (Instituto Universitario de Democracia Paz y Seguridad 2013) making it the most

violent nation in the world (United Nations 2013) Likewise Hondurasrsquo second largest city San

Pedro Sula has the highest municipal level homicide rate in the world with 1290 homicides

(Instituto Universitario de Democracia Paz y Seguridad 2013) or about 174 per 100000

population (United Nations 2013) In comparison the average homicide rate across the globe is

about 62 per 100000 and the average homicide rate in Central America is about 27 per

100000 (United Nations 2013)22

Much of the discussion about the causes of Hondurasrsquo high homicide rate has focused on its

relationship with international drug trafficking routes gangs and conflict between crime

groups and the government and government instability Estimates of gang involvement very

widely but some have suggested that there are between 12000 (Seelke 2012) to 36000

(Ratcliffe et al 2014) gang memberrsquos in Honduras who typically belong to one of two gangs

MS-13 and 18th Street These gangs are said to be less organized than their counter parts in El

Salvador but are said to be just as involved in extortion and intimidation and perhaps more

involved in drug trafficking because of their stronger linkages with Mexican drug cartels

(Wilkinson 2013)

22 An unusual characteristic of the homicide problem in Honduras is the age of victims Typically in the Western Hemisphere homicide victims are aged 15 to 29 In Honduras however those 30 to 44 have the highest rate of violent victimization For example 1 out of 280 males 30 to 44 years old are the victim of homicide compared to 1 out of 360 males 15 to 29 years old (United Nations 2013) These findings by themselves are suggestive of a chronic gang problem (Spergel 1995) Honduras also stands out in the Western Hemisphere in the proportion of its homicides that involve a firearm In 2012 about 84 percent of the homicides involved a firearm (11) The proportion of homicides that involve a firearm appears to be increasing as well In 2008 79 of homicides involved a firearm compared to 81 in 2009 83 in 2010 (United Nations 2013)

71

Over the past decade the nation has responded with ldquoMano Durardquo (ie iron fist or heavy hand)

The new legislation provided the police with more authority to stop search and detain gang

members The new legislation also permitted the courts to sentence gang members to prison

for 12 years for simply being a member of a gang and allowed the courts to sentence

individuals to even longer prison terms for gang related incidents Concomitantly the military

joined the effort to fight gangs by patrolling neighborhoods along side the police While the

public and media strongly supported the shift in national policy toward Mano Dura much of

the evidence suggests that these legislative and policy changes were not effective as the

number of homicides continued to escalate Some suggest that its lack of success was because

gang members who were arrested were released due to of lack of evidence or those who went

to prison if they were not a gang member before entering prison joined a gang Other critics

point out that the heavy handed approach by the government led to loss in the rule of law as

vigilantes engaged in extra-judicial violence against gang members (Seelke 2012) Still others

said that the legislation and policies never really had a chance of working because of the

general lack of effectiveness of the police and courts and the wide spread corruption

throughout the criminal justice system (Zilberg 2011)

As a consequence of the above policymakers and citizens voiced optimism about the possibility

of a truce between gangs after initial results in El Salvador suggested the strategy might be

effective (Villiers-Negroponte 2013) Honduran church leaders and the Organization of

American States (OAS) began to develop a strategy to implement a similar type of truce in

Honduras and the President offered his personal support in their efforts (Arce 2013) In this

case study we examine the processes that lead to the Honduras gang truce and the nationwide

impact of the truce on homicides In the below section we discuss the major stakeholders who

participated in the truce processes leading up to the truce and the establishment of the truce

Key stakeholders

The primary facilitator for the truce process in Honduras was Archbishop Roacutemulo Emiliani who

had earlier served as the Assistant Bishop of the Dioceses of San Pedro Sula and who received

72

support from the Catholic Church to pursue the truce (Bosworth 2013) Prior to the

negotiations he was well known for his work which attempted to establish peace between the

gangs and his advocacy for prison reform and social reintegration programs for gang members

(The Daily Herald 2013) From the onset Monsignor Emiliani proceeded cautiously to ensure

reasonable expectations among the public and policymakers He maintained publically that ldquohe

didnacutet want to be a salesman of false promises about what was going to occur in the future the

things that they do are unpredictable but we expect to have a declaration of reconciliation

principles with societyrdquo (El Mundo 2013) Additionally he wanted to set reasonable

expectations because he knew that it would be a ldquoslow painful and draining processrdquo (El Nuevo

Siglo 2013) and that ldquoWhat is coming is difficult It is not easy It is complicatedrdquo (Castillo

2013)

As in El Salvador the Organization of American States (OAS) played a major role in facilitating

the peace process alongside Monsignor Emiliani Adam Blackwell served as the Secretary of

Multidimensional Security for the OAS and represented Canada on the Honduras Security

Reform Commission (Willcocks 2014) His participation in the mediation process was requested

by Honduran gang members who were in prison They requested that the OAS help broker a

peace agreement with the Honduran government and to help identify resources that would

assist gang members to obtain legitimate jobs (Associated Press 2013) The OAS together with

the Catholic Church served as a ldquobridgerdquo between the executive branch of the government and

the two gangs Additionally two of the mediators (ie Salvadoran Army officer and Police

Chaplain Monsigor Colindres and former Salvadorian congressman Mijango) who helped broker

the truce in El Salvador provided additional support to Monsignor Emiliani and Secretary

Blackwell They traveled to Honduras to present their experiences with the gang truce in El

Salvador and to convey that a gang truce is a promising and legitimate strategy for addressing

gang violence (Associated Press 2013)

Gang leaders of the two primary gangs in Honduras (MS 13 and 18th Street) also participated

extensively in the negotiation process It was stated that they had become weary of the violent

conflict and understood that a truce would be beneficial to the Honduran people (Servellon

73

2013) From the beginning however a number of the critics of the truce argued that Honduran

gangs did not have the capacity to control street level violence They characterized the

Honduran gangs as having less organizational leadership (Bossworth 2013) less control over

turf (LatinNews Daily Report 2013) and being more organizationally fractured (Dudley 2013)

than MS 13 in El Salvador

At the time that discussions about the possibility of a gang truce began President Porfirio Lobo

Sosa was publically supportive of the Catholic Church and OAS negotiating with the gangs

Media reports quoted the President saying I am ldquoprepared to do what ever is necessaryrdquo to

support the mediators (Phillips 2013) ldquoWe have to look for anything thatrsquos an alternative to

violencehellipOn the part of the government we are open to any process that can lower violencerdquo

(Associated Press 2013) and that he had given ldquohis blessing to Emillanirsquos efforts to broker peace

between the gangshelliprdquo (The Daily Herald 2013) However in November 2013 after a general

election the new president Juan Orlando Hernaacutendez through his recently appointed Vice

Minister of Security declared that the government would no longer support the truce process

with the gangs (El Heraldo 2014) Since then the Government of Honduras has not mentioned

the peace process that was initiated in May 2013

Truce making process

It is important to note that prior to the announcement of the gang truce a number of key

stakeholders were somewhat skeptical about its possibility On the one hand some suggested

that a gang truce had been attempted in the past with no success For example one

stakeholder commented to an international media outlet that ldquoEveryone here agrees itrsquos a

positive step forward but people are cautiously optimistic because in 2005 these two gangs

had another peace treaty with each other Now that treaty was very tentative it only lasted

less than two monthsrdquo (Al Jazeera 28 May 2013) On the other hand as noted above other

stakeholders believed that the local gangs did not have enough organizational leadership to

change the behavior of gang members to reduce violence (Bosworth 2013) They argued that

even if gang leaders wanted a gang truce there was no way of enforcing it on the streets

74

Several months prior to the announcement of the gang truce Carlos Mojica Lechuga an 18th

Street Salvadorian gang leader publically stated that representatives of MS13 and 18th Street

in Honduras spoke with several gang truce key stakeholders in El Salvador for the purpose of

replicating the truce in Honduras Tellingly in reflection of the visit Mojica noted that the

advantage of a gang truce is that it formally recognizes Honduran MS13 and 18th Street leaders

as important political persons within the nation He also noted that Honduran gang leaders

have historically been treated poorly and that a gang truce holds the potential for

demonstrating the political power of each of the Honduran gangs (Villiers Negroponte 2013)

The negotiators used different language to describe the early days of the truce Specifically

they mentioned that there had been a consultative process with the gang leaderships so they

were in a process like lighting23 Prior to the truce media sources mentioned that the gang

leaders were offering to stop violence and to not recruit more youth into the gangs24 The gang

leaders also spoke about the ldquopersecutionrdquo they and their family members had suffered during

the previous years highlighting that they had been prohibited social opportunities offered to

others in society (Arce 2013)

Leading up to the negotiations the leadership of both gangs expressed their interest in three

goals 1) lowering violence and crime 2) reconciliation with God society and the government

and 3) helping to improve the social conditions of their communities Although there was not

written documentation on the exact agreement between the parties one MS-13 leader

affirmed that the pact would include all violence (El Comercio 2013) However when talking

about sensitive topics such as extortions which is one of the main sources of income for the

gangs the gang leader said that ldquowould be taken up at a later daterdquo (ABC Internacional 2013)

Leaders of 18th street made similar general statements about ending violence but they were

more specific about their demands One of the 18th Street leaders stated that ldquowhat we want is

23 El Universal 2013 - httpwwweluniversalcominternacional130531obispo-hondureno-descarta-tregua-entre-las-pandillas 24 (Garcia 2013 -httpwwwlaprensahncspmediapoolsitesLaPrensaHondurasSanPedroSulastorycspcid=338546ampsid=276ampfid=98-)

75

to have a dialog with any commission appointed by President Porfirio Lobo and we are sure

that the situation in Honduras will begin to changerdquo (El Nuevo Siglo 2013)

Establishing the gang truce

On May 28th 2013 with public declarations from leaders of both gangs the gang truce was

announced From the beginning of the process the role of the government in the truce was

unclear (La Prensa 2013) Likewise there was little discussion about the exact nature of the

agreement the terms in which gang members would abide and any benefits that would be

made available to those who participated in the truce For example as one stakeholder

indicated the government never decisively considered viable proposals to give the members of

the gangs any opportunities It is important to note that none of the parties signed a formal

commitment and neither MS-13 nor 18th Street signed any type of ceasefire agreement The

gang leaders were in separate locations and were never in direct contact with each other during

the announcement That is both gangs seemingly agreed to the gang truce without ever

talking to each other The ldquopeace processrdquo was publicly announced on May 28 2013 through

ldquojoint but separaterdquo declarations made by the leaders from both gangs imprisoned in the San

Pedro Sula prison (National Penitentiary SPS) (The Daily Herald 2013) The national and

international media widely covered the declarations

The MS-13 leaders said they would not commit any more homicides or any other types of

crimes They ensured that this was an ldquoimmediaterdquo order and would be effective throughout

the country They emphasized ldquoall of the boys know what they have to do starting todayrdquo

(Pachico 2013) The leadership of 18th Street declared that they would stop violence and other

criminal activities but also indicated that the government would have to ldquolisten to themrdquo

Little research has examined whether the gang truce in Honduras ever impacted violence in the

nation Instead anecdotes have been used to portray its effectiveness One facilitator for

example indicated ldquoin Honduras the dialog with the gangs has been positive however the sad

thing in Honduras is that the two main gangs have not accepted a truce between them they

just haven`t accepted it as yetrdquo (La Prensa 2013) despite the fact that ndash in their own words ndash

76

ldquothey do want to hold a dialog with society with the government and with the policerdquo

Conversely gang leaders declared that the truce had been effective For example a leader of

18th Street noted that ldquohellipIt has already done its part telling members in the areas the gang

controls to stop the violence and crimehellip[estimating] crime had already dropped 80 percent in

those areasrdquo (Associated Press June 17 2013) Similarly a member of MS13 estimated that

violence in MS13 controlled areas declined by 45 percent (Associated Press June 17 2013) As

a symbolic gesture of the impact of the truce MS13 leaders also noted that as a gesture of good

will they made and delivered 60 beds for a nursery home in San Pedro Sula (Associated Press

June 17 2013)

Methods

For the present case study we used a pre-test post-test longitudinal quasi experimental design

Data from the 2001 Honduran Census was obtained from the National Institute of Statistics

(INE) These data provided municipal level measures of number of residents population density

per kilometer percent of population who moved in from another municipality ethnicity

percent urban number of residents immigrated to the United States percent female headed

households percent unemployed age composition income percent of households rented and

education level In addition population projections for the years 2005 2010 and 2014 were

also obtained from the National Institute of Statistics (INE) The population levels for the total

population as well as the percentage of residents in a municipality that rural and percentage of

residents who are female were linearly interpolated for the intervening years Examination of

the observed levels indicated that growth was linear overall and so we feel confident that our

linear interpolations are good approximations Second we used municipal level homicide data

by month and year for the period May 2012 through July 2014 These data were provided by

the Honduran National Police through the Honduran US Embassy Both datasets were

merged for the present analysis

Measures

77

The dependent variable for the Honduran case study is the monthly homicide rate which was

calculated by dividing the number of homicides in each municipality by its population and

multiplying this figure by 100000 We examined change by comparing the homicide data 13

months prior to the gang truce to the homicide data 14 months following the truce in each of

the nationrsquos 298 municipalities As presented in Exhibit 21 there were a total of 7910 homicides

over the study period with each municipality averaging 183 homicides (sd=837) A trend

analysis showing the monthly number of homicides on a national level prior to and following

the gang truce is presented in Exhibit 22 It shows that nationally the homicide rate gradually

declined over the study period

Exhibit 21 Summary Statistics

Pre-Truce (n=3809) Post-Truce (n=4101) Total (n=7910)

Mean SD Mean SD Mean SD

Homicide 200 900 166 773 183 837

Population 2874492 8674659 2939557 8870267 2908226 8776124

Homicide rate 521 953 413 814 465 885

Density 9926 10878 9928 10878 9927 10877

Born in the same municipality 8211 1286 8211 1286 8211 1286

Other population group 8545 2390 8545 239 8545 2390

Percent rural 7926 2732 7899 2762 7912 2748

Living in another country 331 650 331 65 331 650

Socioeconomic status 005 097 -002 102 001 100

Exhibit 22 Homicide rate in Honduras by month

78

We used census data to control for several community-level structural factors Principal

components analysis was used to reduce some of the data into a summary measure Exhibit 23

presents the results of the component loadings One component was extracted that we labeled

socio-economic status (SES) This component exhibited high loadings for percent female

headed household percent unemployment and percent completing primary education

Population was used as our exposure variable and population density (per 1 km) residential

stability percentage of the population that is rural percentage of the population that is

indigenous (verify) and the number of residents immigrated to the United States served as our

control variables Population density and outmigration were logged to address skewness in

these two measures

Exhibit 23 Factor Analysis for Socioeconomic Status

Eigenvalue 236093

Variables Factor Loadings

Female-Headed Households 07977

Unemployed 09269

Primary Education 09303

Findings

The results of our t-test analysis are shown in Exhibit 24 It shows that there was a significant

decline in the homicide rate at the municipal level before and after the gang truce The

homicide rate prior to the truce was 697 per 100000 population and following the truce the

homicide rate was 566 homicides per 100000

79

Exhibit 24 T-test on National Data by Month (n= 27 Months)

Homicide Rate Pre Truce Post Truce Change

Mean 6972346 5663592 -1308754

SE 01677361 01151668 02009224

plt 001

Next we examined the effect of the truce through ARIMA regressions The first model in Exhibit

25 does not include an effect of the temporal trend In this model we again find that homicides

declined significantly in the period following the truce We then employed the ARIMA model

again but this time included a variable (month) to control for the temporal trends in the data

Model 2 in Exhibit 25 shows that when we controlled for temporal trends the impact of the

truce we observed was no longer significant Our findings suggest that homicide did not decline

as a consequence of the gang truce but instead the decline in homicides was part of a longer

term historical trend that was independent of the truce

Exhibit 25 Results from the ARIMA Models (n= 27 Months)

Homicide Rate Model 1 Model 2

Effect SE Sig Effect SE Sig

Truce -1309 0350 -0906 0595

m2 -0623 0481

m3 -0153 1447

m4 -0074 0607

m5 -0104 0450

m6 -0050 0466

m7 -0002 0516

m8 0003 0404

m9 0024 1633

m10 -0396 0522

m11 -0754 2424

m12 0837 0366

Time -0030 0047

Intercept 6973 0254 7258 0472

Autocorrelation Coefficient 0010 0408 0111 0249

plt 001

80

Last we used a fixed effects multi-level negative binomial regression to predict the homicide

rate with population as an exposure variable and controlled for the socio-demographic

characteristics of each community The results are presented in Exhibit 26 The only significant

variable in the analysis was the impact of municipal level population stability Specifically we

found that homicide rates increased in areas where residential mobility was high Once again

our analysis showed that time had a negative effect indicating that homicides were decreasing

in all areas over the study period The gang truce itself was unrelated to the decline in

homicides

Exhibit 26 Results from the Multilevel Negative Binomial Models (n= 27 Months)

Homicide Rate Model 1 Model 2

Effect (SE) Sig Effect (SE) Sig

Fixed Effects

Homicide Rate 0004 0004

(0002)

Truce -0119 -0129

(0073) (0072)

Time -0008 -0007

(0005) (0005)

Month 1 0024 0024

(0059) (0059)

Month 2 -0107 -0107

(0062) (0062)

Month 3 -0041 -0041

(0063) (0063)

Month 4 -0033 -0033

(0065) (0065)

Month 5 -0002 -0002

(0067) (0067)

Month 6 (reference)

Month 7 0027 0027

(0051) (0051)

Month 8 0023 0023

(0057) (0057)

Month 9 0036 0036

81

(0056) (0056)

Month 10 -0051 -0051

(0057) (0057)

Month 11 -0031 -0031

(0058) (0058)

Month 12 0187 0187

(0056) (0056)

Born in the same municipality -0013

(0003)

Pecent rural -0003

(0002)

Socioeconomic status 0029

(0044)

Percent dominant population 0001

(0002)

Density (ln) -0010

(0050)

Living in another country (ln) 0001

(0061)

Intercept -10042 -10051

(0061) (0058)

ln(Alpha) -2368 -2367

(0148) (0148)

Random Effects

Var(Truce Effect) 0066 0066

(0024) (0025)

Var(Intercept) 0388 0309

(0049) (0042)

Cov(truce effect intercept) -0024 -0006

(0027) (0025)

plt 001

82

Conclusions

Violent crime in Honduras is widespread With a homicide rate of about 86 per 100000

population Honduras is one of the most violent nations in the world (United Nations 2013)

This compared to an average homicide rate around the world of 62 per 100000 people and

about 27 per 100000 in Central America (United Nations 2013) Given the high rate of violence

in Honduras it is clear that new and innovative ways of reducing violence should be considered

This case study gave an overview of the implementation and impact of the gang truce

experience in Honduras in 2013

The goal of the truce was to significantly reduce the number of homicides In order to assess

the impact of the gang truce in Honduras a pre-test post-test longitudinal quasi-experimental

design was used Census and homicide data were merged at the municipal level to asses

whether the truce had an impact on homicides controlling for population characteristics and

the natural trend in violent crime

Overall the findings suggest that while the homicide rate in Honduras was on a slight

downward trend the gang truce itself was unrelated to any homicide reductions That is the

gang truce had no measurable impact on homicides in Honduras Given the drastic reductions

achieved in El Salvador and the fact that Hondurasrsquo gang truce was a replication of El

Salvadorrsquos the following discussion will examine two important differences between the two

countries gang truces First the implementation of the truce in Honduras did not appear to

obtain trust between all parties involved and did not achieve any notable short-term

deliverables That is the implementation was not robust Second some suggest that the gangs

in Honduras do not have the organizational sophistication to be able to control their members

on the streets rendering them incapable of carrying out any truce agreements

First the implementation of the truce in Honduras did not appear robust The communication

between the parties was weak and none of the parties completed any significant actions as part

of the truce The Catholic Church and OAS served as the bridge between the two main gangs in

Honduras MS13 and 18th Street and the executive branch of the government The ldquopeace

83

processrdquo as it was called in Honduras was initiated in May 2013 After the general election in

November 2013 the newly appointed administration declared that it would no longer support

the truce process with the gangs As a result the truce was short-lived and the governmental

support for the effort shifted with the change in leadership

Though the negotiations were largely based on the good will of a well-respected negotiator

few tangible incentives were offered during the process The negotiations from the beginning

involved discussions about large scale social programming Given the short time frame these

goals in hindsight were unrealistic Naturally there was little trust between the parties at the

beginning of the negotiations and without any quick tangible deliverables from either side the

truce never really materialized

The second challenge to implementing a successful gang truce in Honduras might have been

related to the nature of the gangs themselves It was unclear whether the gangs possessed the

level of cohesion and hierarchical leadership required to make some measures feasible If the

gangs do not have the organizational capacity to control their members on the streets any

agreements that come from the negotiations would be difficult to implement In the earlier

case study of the Salvadorian experience it was suggested that the successes in that country

were related to one of the gangrsquos organizational capacity to impose control of its members The

gangs in Honduras might be different There is at least some evidence that MS 13 in Honduras

might have less organizational leadership (Bossworth 2013) less control over turf (Latin News

Daily Report 2013) and have been more organizationally fractured (Dudley 2013) than their

counterparts in El Salvador The gangrsquos ability to operate as an efficient organization can greatly

impact the outcome of the truce process In general we know that gangs do not have very high

levels of organizational sophistication (Decker Katz and Webb 2008 Decker Bynum Weisel

1998) It might be that gangs in Honduras are more the norm in terms of organizational

capacity

In summary the 2012 truce negotiations in Honduras did not produce any measurable

reductions in the homicide rate The gangs wanted to speak to Honduran society and they even

84

preferred to speak with the governmental authorities but they never interacted with either

The negotiation process seemed to end as quickly as it started The gangs did not deliver with

lower rates of violence and the government did not provide social programs It would be safe to

say that a robust gang truce did not materialize in Honduras The implementation of the truce

seemed to struggle for two primary reasons First the mediators were not able to accomplish

any quick wins to build trust between the parties involved The commitment level on all sides

was not clear throughout the process Second it is not clear whether the gangs in Honduras

have the organizational capacity to control members on the street as would be required to

carry out an effective gang truce In short the 2013 gang truce in Honduras was unsuccessful

85

Conclusions Policy Implications and Recommendations

The purpose of this report was to systematically examine the effectiveness of gang truces Gang

truces have been widely implemented but rarely evaluated Of those gang truces that have

been evaluated little attention has been given to why and how they came into existence In

this report we reviewed prior research on gang truces and presented case studies of gang

truces implemented in El Salvador Jamaica and Honduras for the purpose of understanding

the negotiation processes undertaken with and between gangs and other stakeholders We

were interested in identifying the actors involved in the negotiations the goal(s) of the

negotiations and the strategies employed to carry them out Most importantly however we

wanted to determine whether the gang truce resulted in a reduction in the number of

homicides Each case study offers lessons learned that are unique to their particular

circumstances and when considered together provide direction to policymakers on the

benefits and risks of implementing gang truces

The case studies presented here constitute the most comprehensive evaluations of gang truces

to date Existing documents were used to collect information about the processes associated

with each gang truce Many of these documents included such items as peer reviewed articles

books and reports The majority of these documents were collected over the Internet

requesting documents from those close to the truce and searching library databases Related

the case studies made use of articles obtained from local newspapers The newspaper articles

were not only intended to provide a historical record of the development of each gang truce

but also to provide additional insight into the various external forces that might have impacted

the gang negotiations Because the newspaper serves as a forum for the community to speak

about its concerns newspaper articles also provided a rich source of data on how those in the

community felt about the gang truce Accordingly the newspaper articles offered a different

view of the problem and offered different opinions as to how a gang truce should or should not

be implemented We also conducted a small number of in-depth qualitative interviews with

key informants These data were collected to supplement existing documents and to clarify

issues associated with the negotiation processes This included but was not limited to

86

questions pertaining to identifying the actors involved in the negotiations the goal(s) of the

negotiations and strategies employed to carry out negotiations The interviews were intended

to obtain information from those who possessed first hand knowledge about the gang truce in

each nation

We examined the impact of each truce using official data We first performed a simple t-test

comparing the homicide rates before and after the truce However as discussed above this

technique has limitations The most severe of which is that even if the test was significant it

would be difficult to determine whether the difference was due to the gang truce or a natural

change over time in the outcome We addressed this limitation by using time series models

whereby the homicide rate for the community was modeled as a function of time with truce

period indicators included to measure the effect of the truce net of the temporal trends These

models were estimated with ARIMA techniques Supplemental models were also employed to

examine and control for factors other than the truce that might have impacted homicide over

the study period

Summary of Findings Related to the Implementation of a Gang Truce in the Three Sites

We found that the implementation of gang truces have a number of common characteristics

The first is that in each case a community was experiencing an uncharacteristically high number

of gang related homicides over a fairly lengthy period of time The continued high level of

violence in each case resulted in the community placing strong pressure on the government in

general and the justice system in particular to respond to the problem quickly and effectively

In each case they had first attempted to control gang violence through suppression oriented

strategies and these strategies were found to be ineffective over the intermediate and long

term In turn each communityrsquos inability to exercise traditional informal and formal social

control to decrease levels of violence became self evident to the public and government This

resulted in both the state and community to seek (or participate in) an alternative strategy in

which negotiators would formally andor informally work with gang leaders to establish a truce

that would reduce gang homicide

87

Key stakeholders involved in the negotiation and establishment of each gang truce were fairly

similar In each of the cases examined the gang leaders of the largest and most violently

involved gangs were willing to consider participating in negotiations that could lead to a truce

In each of these cases it was clear that the gangs not only sought to collaborate with the

negotiators for the purpose of reducing violence but perhaps more importantly were seeking a

means in which to gain greater more positive recognition in the community and to reap some

form of benefit to themselves their members and possibly their community In each case

while not always formally involved government officials were at a minimum made aware of

negotiations and in some cases solicited the assistance of third partyrsquos to broker an agreement

between stakeholders In each case it was at least implicitly understood that the government

would ldquolistenrdquo to the gang leaderrsquos expectations of the government and what theymdashthe gang

leaders--had to offer in exchange We found that when the government was no longer willing

to ldquolistenrdquo to or collaborate with negotiators the truce processes ended abruptly Negotiators

were typically comprised of a very small group (ie 2-3) of individuals who were perceived to

be ldquohonest brokersrdquo In El Salvador and Honduras this included a high ranking Catholic Church

official a leader from an international diplomatic organization (ie OAS) and other neutral

parties In Jamaica this included a quasi-governmental organization that had been established

for the purpose of brokering negotiations between gangs to reduce violence and the local

university which had access to staff who were perceived to be neutral but had an interest in

reducing violence due to its proximity to the university

The strategies used to execute each gang truce were similar but yet importantly different They

were similar in that each involved a team of negotiators working to identify common goals to

be achieved and identifying tangibles that could be delivered to the gang leaders gang

members and their community in exchange for the gang achieving their stated goals They

were different however in terms of the structure of the delivery of each parties promise to the

other In Honduras and Jamaica it appears that gang leaders committed to reducing gang

violence in exchange for general promises made by the negotiators for example that

substantial public works programs would be implemented for the goal of reducing

unemployment among gang members and the community In both of these cases it required

88

the government to develop and deploy large scale social programming in a very quick period of

timemdashsomething that neither government had a strong record of demonstrating In El Salvador

negotiators employed a strategy of the gang leaders promising to deliver immediate changes in

gang member behavior for immediate administratively natured changes by the government

For example in exchange for a reduction in gang violence the government agreed to relocate

imprisoned gang leaders to less restrictive prisons and provide them some privileges Following

the successful execution of the first part of the truce which resulted in near term success for

both parties they began to negotiate issues that would take a longer period of time for the

gangs and government to deliver Our findings suggested that some promised deliverables need

to be easily and quickly delivered early in the process so that trust increases between both

parties Stakeholders only have a brief period of time to provide promised benefits before trust

is lost and that tangible benefits need to be delivered in weeks or months not years

Summary of Findings Related to the Impact of Gang Truces in the Three Sites

Our findings suggest that El Salvadorrsquos gang truce had a significant and dramatic impact on the

homicide rate The mean number of monthly homicides declined from about 354 prior to the

truce to about 218 following the truce for a net decrease of about 136 homicides per month

Our forecast showed that between March 2012 and June 2014 the truce had saved about 5501

lives However there was no significant change between the pre-truce and post-truce periods

in the number of thefts extortions robberies rapes and auto theftsrobberies We also found

that the gang truce did not result in a homogenous decline in violence across municipalities

About 61 percent of municipalities experienced a decline in homicides but the decline in

violence varied substantially between municipalities We examined this issue further by parsing

out the relative influence of the number of MS13 and 18th Street gang members on the street

and in prison from each municipality Our analyses indicated that following the truce the

number of MS13 and 18th Street gang members on the street in a municipality was not

significantly related to a decline in homicide but the number of imprisoned MS13 and 18th

Street gang members was associated with a significant change in homicides following the gang

truce In particular the number of imprisoned MS13 gang members was associated with a

89

significant decline in homicides following the gang truce and the number of imprisoned 18th

Street members was associated with a significant increase in homicides following the truce

In Jamaica our initial findings showed that the gang truce might be an effective mechanism for

reducing violence Bivariate analyses showed a significant decline in homicides immediately

after the truce was implemented This explains the work previously published by policymakers

researchers and news reporters Upon further examination of the data however comparing

change in the target and comparison areas and accounting for temporal trends we found that

the decline in homicide was part of a larger nationwide decline in violence and that the gang

truce was not responsible for the decline The only significant effect that we uncovered was the

possibility that homicides were displaced outside the target area for a brief period of time but

then returned to normal

Our findings from Honduras told a similar story as Jamaica Initial analysis showed that the

number of homicides on average declined across municipalities following the gang truce

Specifically the mean number of homicides declined by 13 per 100000 population with 687

homicides per 100000 population on average occurring in each municipality prior to the truce

and 566 homicides per 100000 population on average occurring in each municipality after

the truce However after we examined the effect of the truce through the ARIMA model and

included a variable (month) to control for the temporal trends in the data the impact of the

truce we observed in our bivariate analysis was no longer significant Our findings as in

Jamaica suggested that the decline in homicides was not the consequence of the gang truce

but instead the decline in homicides was part of a long term decline in homicides due to

exogenous factors

The Potential Benefits and Consequences of a Gang Truce

Over the last several years there have been a number of naturally occurring experiments

involving gang truces in a variety of nations in various regions of the world Findings from

evaluations of gang truces are mixed As noted above in El Salvador the gang truce could be

characterized as highly effective at least for the two years following the truce It is worth

mentioning that even after the truce breakup homicides rates while above truce levels

90

continue slightly below pre-truce levels In Jamaica and Honduras the gang truce had no impact

on violence In Los Angeles and Trinidad there was evidence that violence decreased for at

least ninety days but then increased substantially beyond those rates observed prior to the

gang truce (see the introduction section of this report for this discussion) As a consequence it

appears that the potential for long term consequences might out weigh the potential for short

term benefits Only one study site(El Salvador) demonstrated a truce having a substantial and

long term impact on violence Others conversely demonstrated the truce had no impact or

increased violence over the long term In fact a number of scholars have noted that gang truces

are likely to result in a boomerang effect with gang violence increasing over the long run

because of enhanced cohesion within the gang (Klein 1995) Maguire (2013) notes that when

government officials negotiate a truce with gangs they might ldquoinadvertently be acknowledging

gangs as legitimate social entitiesrdquo (p 11) This in itself might increase cohesion among gangs

which has been found to be associated with increased levels of criminality (Decker et al 2008

Klein 1971 Maguire 2013)

Further research is needed to examine how gang truces might impact group cohesion and if it

does whether this in turn results in greater violence Gang truces convey the well-intentioned

image that violence has been addressed and policymakers are doing something about the

problem but researchers need to better understand the probability of a gang truce reducing

violence increasing violence or having no impact This will better position policymakers to

understand the relative risks associated with these types of interventions

Our findings also suggest that while gang truces could be an effective intervention in areas

where gangs are highly structured and organized such as El Salvador they could be counter-

productive in areas where gangs are not as structured and organized Because the vast majority

of street gangs are not well organized (Klein 1995 Spergel 1995) the utility of a gang truce in

reducing violence might be limited Our findings coupled with prior research suggest that gang

interventions need to be tailored to the nature of the gang and its members or it risks

increasing gang violence

Final Thoughts

91

Our analysis suggests that gang truces should only be used as a means of last resort and then

only under certain conditions Given the risks associated with a gang truce communities with

high levels or at least modest levels of formal social control should rely on other more

promising gang control strategies Only when the state has limited or greatly reduced capacity

for social control should a truce be considered Concomitantly a gang truce should only be

considered when a community is experiencing a substantial amounts of gang violence

Communities that are experiencing minimal to modest amount of gang violence may have more

to lose from the establishment of a gang truce than they have to gain Additionally our

findings suggest that a gang truce might only be feasible when gangs are sufficiently organized

to the extent that they have the capacity to regulate memberrsquos behavior In other words gang

leaders must have the ability to reduce their memberrsquos involvement in violence for a gang truce

to work Our findings more concisely suggest that gang truces should only be considered when

there are a great number of gang homicides the state has limited capacity to address the

problem and gang leaders have enough informal social control over their members that they

themselves can substantially control the levels of violence in their community by regulating

their memberrsquos behavior

Recommendations

Gang truces are conjunctural strategies States who suffer from gang-related violence must

establish permanent public policies for crime control and prevention A government that

considers implementing a gang truce should be aware that it cannot become the center

strategy of its public policy for citizen safety

Gang truces should only be used as a means of last resort and then only under certain

conditions Stakeholders must determine whether a process of dialog or negotiation with gangs

is legal ethical and feasible

Stakeholders must anticipate demands that are likely to arise and their response options Some

demands may be easily met such as improved prison conditions Others are much more

difficult and amorphous such as community development through more integrated violence

92

prevention programs (such as those implemented by SolucionES in El Salvador) local economic

development programs or economic reinsertion of ex-gang members

Stakeholders should incorporate immediately achievable and demonstrable deliverables Long-

term goals and promises are unlikely to create the trust needed to sustain a gang truce

Stakeholders must first determine the position in which they are negotiating the incentives that

are possible to deliver and the boundaries and limits they face Gangs are mostly likely to trust

representatives from NGOs community-based organizations and members of the faith based

community as brokers because they are considered more reliably neutral advocates for peace

They need to understand the capacity of the government to deliver promises in a timely

manner

Governments have to make a choice about the visibility and transparency of its participation

This decision needs to be made in the context of the national and local laws the publicrsquos

expectations of transparency and patterns of practices of the past

Governments must be strategic in their support for truce initiatives Some donor funded

programs run by the government prohibit gang member participation and if the government

does not receive approval from the donor it may risk the donor withdrawing its sponsorship of

the program

Governments must ensure an inter-institutional coordination for the management of truces to

avoid the responsibility to be of a single government institution It is necessary to generate or

collect reliable and pertinent data that can be used to analyze and assess the process

It is necessary to implement an effective monitoring system of the truce process similar to that

used in the full report as it can help parties understand what is working and who is delivering

on their promises More specifically the monitoring and truce management system should be

able to identify truce violations and be prepared to respond through the use of legal and

effective practices if stakeholders do not comply

93

Finally it is necessary to develop evaluations of gang truces and monitoring programs and

support violence prevention activities local economic development activities and pilot

programs to support the reinsertion of ex-gang members into society Clearly national

governments municipal governments NGOs and community-based organizations need

increased capacity and resources to discourage the growth of gangs among at-risk youth It is

therefore increasingly important to create economic opportunities for gang members willing to

leave the gangs and find other legal employment Developing and sustaining those

opportunities in nations with high incidences of poverty will require significant international

funding

94

References

Anderson Elijah 1999 Code of the Street New York Norton

mdash 1998 ldquoThe Social Ecology of Youth Violencerdquo In Crime and Justice vol 24 edited by M

Tonry 65-104

mdash 1990 Street Wise Chicago University of Chicago Press

Arce A (2013 May 28) Honduras gangs declare truce ask talks with govt Associated Press

The Big Story Retrieved from

httpbigstoryaporgarticlehonduras-gangs-declare-truce-seek-talk-govt

Archibold Randal 2012 ldquoGangsrsquo Truce Buys El Salvador a Tenuous Peacerdquo The New York Times

Originally published online on August 27 2012

Axelrod R (1986) ldquoAn Evolutionary Approach to Normsrdquo American political science review

80(04) 1095-1111

Ayala Edgard 2012 Gangs Back Plan for Violence-Free Districts in El Salvador

httpwwwipsnewsnet201212gangs-back-plan-for-violence-free-districts-in-el-

salvador Found on November 4 2014

Bakrania Shivit 2013 Policy Responses to Criminal Violence in Latin America and the

Caribbean GSDRC Helpdesk Research Report 934 Birmingham UK GSDRC University

of Birmingham

Bargent James 2013 ldquoRise in Disappearances Feeds Doubts Over El Salvador Trucerdquo August 1

2013 Found at httpwwwinsightcrimeorgnews-briefsrise-in-disappearances-in-el-

salvador-feeds-doubts-over-truce on September 1 2014

Battin Sara R Karl G Hill Robert D Abbot Richard F Catalano and J David Hawkins 1998

ldquoThe Contribution of Gang Membership to Delinquency Beyond Delinquent Friendsrdquo

Criminology 36 93-115

Block Richard 2000 ldquoGang Activity and Overall Levels Of Crimerdquo Journal of Quantitative

Criminology 16 (3) 369-83